⚡Megatron LLM Hub
全部
Website

rss-website

头条2

PHP 位于大量 Web 服务入口,漏洞编号连着 CISA 通告,未打补丁即暴露。

Oct 24, 2026 Live Exploitation for PHP2Shell (CVE-2026-109618, CVE-2026-109619): SafeBreach Coverage for the CISA PHP Alert Learn how PHP2Shell works, which PHP versions are affected, and how SafeBreach goes beyond finding vulnerable web servers by running a live PHP2Shell exploit to test whether your patches and security controls stop the attack. Summary Two critical PHP flaws: PHP2Shell is the name for CVE-2026-109618 and CVE-2026-109619. These are memory-safety vulnerabilities in PHP’s core request handling, and they let an unauthenticated attacker run code as the web worker account. Both are rated 9.2 (Critical) under CVSS v4.0. CISA urges immediate patching: On October 21, 2026, CISA urged organizations to update to PHP 8.5.12, 8.4.27, 8.3.36, or 8.2.35. PHP 8.1 and older receive no fixes from the PHP project. A version check alone can’t tell you if you’re exploitable: The flaws are present in every unpatched PHP installation serving web requests, including PHP-FPM, Apache with mod_php, php-cgi, FrankenPHP, and LiteSpeed. Whether an attacker can exploit them depends on the server setup and PHP configuration. The PHP project has confirmed remote code execution on Apache with mod_php, where the exploit fires before any application code runs. Successful exploits are quiet, so controls must be tested, not assumed: CrowdStrike found that a successful PHP2Shell exploit returns a normal HTTP 200 response and leaves application logs clean. Detection has to focus on what follows, like PHP workers spawning shells. SafeBreach coverage validates PHP2Shell with a live exploit: SafeBreach now runs a working PHP2Shell exploit, validated and integrated by SafeBreach Labs, to test whether patches actually closed the flaw and whether endpoint and network controls detect and stop exploitation. To the best of our knowledge, it is the first live PHP2Shell exploitation simulation from a CTEM platform. A second simulation identifies web servers running an affected PHP release. Author: Noam Sagiv, Product Manager On October 21, 2026, the Cybersecurity and Infrastructure Security Agency (CISA) released an alert titled “Critical Vulnerabilities in PHP Require Immediate Patching” . The alert warns that CVE-2026-109618 and CVE-2026-109619 allow unauthenticated remote code execution and affect all PHP installations serving web requests. It also urges organizations to update to PHP 8.5.12, 8.4.27, 8.3.36, or 8.2.35 immediately. CISA also advises organizations that cannot update right away, or that still run PHP 8.1 or earlier, to take PHP-based systems offline until they are patched or updated. The alert follows the PHP project’s own security advisory , published alongside the fixed releases, which rates the risk highest for Apache with mod_php. Cyber agencies in Australia , Canada , and the United Kingdom issued parallel warnings. What is PHP2Shell? PHP2Shell is the name for two critical vulnerabilities in PHP’s core request handling, CVE-2026-109618 and CVE-2026-109619, disclosed on October 21, 2026. Both can let an unauthenticated attacker execute code on a vulnerable PHP web server with crafted HTTP requests, and both are rated 9.2 (Critical) under CVSS v4.0. The CVE records credit researchers at Calif, OpenAI, and Wiz with finding and reporting them. The name is not the PHP project’s own, but it describes the outcome accurately: a crafted HTTP request becomes a shell on the web server. What sets PHP2Shell apart is where it sits. Most web attacks depend on a flaw in the application—a vulnerable plugin, an unsafe upload handler. PHP2Shell sits beneath all of that, in the runtime that parses each request. On the mod_php path that the PHP project has demonstrated, a fully patched application offers no protection because the vulnerable code runs first. VirusTotal fingerprints roughly 5.1 million internet-facing IP addresses running PHP, and PHP is often embedded inside other software, where it may not appear in an asset inventory at all. The PHP project announced on October 20 that fixes would ship the next day and released them on October 21 at 14:00 UTC. Details had already leaked through a now-deleted forum post, including claims that the flaws scored 10.0; PHP rates both 9.2. Is PHP2Shell being exploited? Reporting on in-the-wild activity is still developing, and the sources do not yet agree: CISA had not observed exploitation as of the alert’s publication. Amazon Threat Intelligence reported exploitation attempts against its honeypots beginning October 17, four days before the patches shipped. Amazon reported broad exploitation attempts that appeared to come from multiple actors, while noting it was unclear whether the attempts would have succeeded against real-world systems. The PHP security team said proof-of-concept exploits published before the patches were not functional. As of October 22, Dataminr reported activity limited to scanning, fingerprinting, and non-working exploits, with no known compromises. Cloudflare had seen millions of PHP2Shell exploit attempts by October 22, according to The Hacker News , including failed attempts to install a variant of the Mirai botnet. Several security firms, including SOCRadar and Wordfence, still report no successful compromises. CrowdStrike developed a working exploit during internal testing, which it has not released, and confirmed that a single anonymous POST request can cause an Apache mod_php worker to launch a shell. SafeBreach Labs has also developed a working PHP2Shell exploit, which powers the live exploitation simulation described in the SafeBreach coverage section below. Amazon’s guidance is the right frame: defenders should not assume the public disclosure date is the earliest date of exploitation. How does CVE-2026-109618 work? PHP uses longjmp() to abandon a request when it runs out of memory. According to the CVE record, some code paths do not account for this and leave request memory uninitialized before performing an allocation. The request cleanup routine then reads or branches on that memory, and multiple out-of-memory strategies can consistently turn the condition into remote code execution. The attack completes within a single request, and CrowdStrike’s analysis places the vulnerable parsing in PHP’s handling of request input, before extensions or scripts run. Configuration matters here. CrowdStrike reports that the URL-encoded path it demonstrated requires PHP’s memory_limit to be lower than roughly four times post_max_size, which PHP’s defaults (128M and 8M) do not meet. The larger upload limits common on sites that accept files—64M against a 128M memory limit, for example—do, and they can be set per directory or per virtual host. How does CVE-2026-109619 work? PHP keeps some global state that temporarily points into per-request memory. In certain scenarios that state is not correctly reset when the request ends, so it continues to reference memory that has already been freed. A later request handled by the same worker can read that stale state before initialization resets it, and the CVE record notes multiple strategies that turn this into remote code execution (RCE). Exploitation takes two steps: one request poisons the worker, and a second triggers the flaw. The PHP project’s confirmed paths run through session upload progress tracking and, on PHP 8.5, fatal error backtraces, though PHP notes these are not the only variants. Long-lived workers, such as Apache with mod_php, carry the most exposure. Which systems are affected by PHP2Shell? The PHP project says the flaws affect all PHP installations serving web requests, including PHP-FPM, Apache with mod_php, php-cgi, FrankenPHP, and LiteSpeed. Being affected doesn’t mean every installation is exploitable. The PHP project has confirmed remote code execution on Apache with mod_php and assesses PHP-FPM, FrankenPHP, and LiteSpeed as likely or possibly susceptible. Configuration matters too: the path CrowdStrike demonstrated doesn’t work under PHP’s default memory settings. Patching is still the only complete fix, because PHP notes that other exploitation variants exist. Fixed releases are available for every supported branch: PHP 8.5: fixed in 8.5.12 PHP 8.4: fixed in 8.4.27 PHP 8.3: fixed in 8.3.36 PHP 8.2: fixed in 8.2.35 (this branch reaches end of life on December 31, 2026) PHP 8.1 and older no longer receive updates from the PHP project, and CrowdStrike reports that the vulnerable code is present across the PHP 7.x and 8.x series. Of the roughly 1.1 million internet-facing IP addresses that disclose their PHP version, VirusTotal counts about 758,000 on end-of-life branches (5.4 through 8.1) and about 272,000 on supported branches. The PHP security team has published minimal source backports for PHP 7.4, 8.0, and 8.1, and some Linux distributions and long-term support vendors provide patched packages for older branches. PHP2Shell TTPs The CISA alert does not map PHP2Shell to MITRE ATT&CK, so the activity below is described by stage rather than by technique ID. Reconnaissance Activity observed since disclosure has been dominated by scanning and fingerprinting, as actors identify internet-facing servers and the PHP versions they run. A server that advertises its PHP version in a response header answers that question for them. Initial Access The attacker sends crafted HTTP requests to an internet-facing PHP server. No authentication, session, or vulnerable application code is required. On the mod_php path the PHP project has demonstrated, the flaw triggers while PHP is still processing the request, before any application code runs. Execution Successful exploitation yields code execution with the privileges of the web worker account. CrowdStrike demonstrated an Apache mod_php worker launching a shell process from a single anonymous POST request, and CISA advises organizations to monitor for PHP worker processes launching shells or downloading tools. Defense Evasion A successful PHP2Shell exploit is quiet. CrowdStrike reports that the worker process does not crash, the server returns a standard HTTP 200 response, and application logs record nothing unusual. The only server-side trace it observed was a PHP memory-limit fatal error ending in “in Unknown on line 0.” CrowdStrike notes that a failed attempt is absorbed by a worker respawn that monitoring treats as routine, so bursts of respawns are worth a closer look. What Follows Initial Access No public reporting yet describes post-exploitation activity from a successful PHP2Shell intrusion. A shell running as the web worker is a foothold, not an endpoint, and defenders should plan for standard web-tier tradecraft: web shells written into the document root, command execution from the web server process, tool downloads, outbound connections to attacker infrastructure, and theft of database credentials and API keys stored in files the web user can read. PHP2Shell IOCs The CISA alert does not publish IOCs, and no authoritative list of malicious hashes, domains, or IP addresses has been released. Amazon reported that exploitation attempts came from commercial anonymization infrastructure, but did not publish the addresses. Detection therefore rests on behavior and logs: the memory-limit fatal error described above (which also appears in benign failures, so it warrants investigation rather than a verdict), bursts of worker respawns, the oversized and malformed request shapes PHP’s interim mitigations are designed to limit, and PHP workers spawning shells. Each is covered under Proactive Threat Monitoring below. SafeBreach Coverage & Playbook Attack Updates PHP2Shell is a vulnerability in a runtime rather than a malware family, so coverage has three jobs: finding PHP servers running an affected release, confirming that patches actually closed the flaw, and validating that endpoint and network controls detect and stop exploitation. To address all three, SafeBreach has added the following new coverage: #11841—Reconnaissance of PHP releases affected by PHP2Shell CVE-2026-109618 CVE-2026-109619. Reads the PHP version a server exposes, compares it with the first fixed release for its branch, and reports whether the server is vulnerable, patched, end-of-life, or unknown. It runs safely, using only the version information the server already exposes. NOTE: This is an Advanced Action and should be scheduled accordingly. #11842—Remote code execution through the PHP session upload progress vulnerability php2shell CVE-2026-109618 CVE-2026-109619. Runs a working PHP2Shell remote code execution exploit, developed by SafeBreach Labs, against PHP servers. It tests both whether a server’s patch is effective and whether endpoint and network security controls detect and stop exploitation as it happens. NOTE: This is an Advanced Action that may disrupt the target service, so run it only against servers you are authorized to test and schedule it accordingly. How to Assess Your Exposure & Test Your Defenses Against PHP2Shell with SafeBreach SafeBreach customers can now assess their PHP2Shell exposure, verify their patches, and validate endpoint and network controls against a live PHP2Shell exploit in multiple ways. Method 1 – SafeBreach Scenarios: Navigate to the SafeBreach Scenarios page and choose PHP2Shell (CVE-2026-109618, CVE-2026-109612). Method 2 – Attack Playbook: From the Attack Playbook, filter by the threat name PHP2Shell (CVE-2026-109618, CVE-2026-109612) or search for the attack ID listed above. Method 3 — Known Threats Series: Open the Known Threats Series report, select the PHP2Shell (CVE-2026-109618, CVE-2026-109612) report, and run the simulations. Additional Advisory Steps Mitigation Strategies Update PHP immediately to 8.5.12, 8.4.27, 8.3.36, or 8.2.35, prioritizing internet-facing systems, as CISA advises. Restart every PHP worker, or use a reload that replaces both the binary and every worker, and confirm the version serving each site rather than the command-line version. Inventory every PHP runtime , including PHP bundled inside content management systems, appliances, and third-party applications, where it may not appear in standard asset lists—a visibility gap CISA calls out in its alert. Resolve unsupported branches. PHP 8.1 and older receive no fixes from the PHP project. Obtain backported packages from your Linux distribution or long-term support vendor, or upgrade to a supported branch. CISA advises taking systems offline that cannot be patched or updated. Treat interim measures as a stopgap only. The PHP project stresses that its interim settings, such as lowering post_max_size, disabling session.upload_progress.enabled, and enforcing request size limits at the web server or proxy, reduce exposure to specific attack paths but do not make an installation safe. See the PHP advisory for the full list and its caveats. Use WAF protection as a bridge, not a fix. CISA notes that a web application firewall may provide some protection until systems are patched. It does not substitute for updating. Check patched systems for prior compromise. Given reports of attempts before disclosure, look for new or modified .php files in web roots and for unexpected cron jobs, processes, user accounts, and outbound connections. If anything looks wrong, rotate every secret the web user could read. Proactive Threat Monitoring Alert on web server processes spawning children: php-fpm, php-cgi, Apache or httpd with mod_php, lsphp, and frankenphp should rarely launch shells, interpreters, or network tools. Hunt PHP error logs for memory-limit fatal errors ending “in Unknown on line 0,” and correlate them with the requests that preceded them. Watch for bursts of worker respawns in PHP-FPM and Apache logs, which can indicate failed exploitation attempts. Flag anomalous request shapes: oversized or deeply nested POST bodies, very high field or multipart part counts, requests without a usable Content-Length, and session upload progress fields on endpoints that do not use sessions. Log outbound connections from the web server user and alert on any destination without a business reason. Stay Ahead with SafeBreach Patching closes the two flaws, but only if the patch is applied everywhere and works as expected. SafeBreach customers can run both new PHP2Shell simulations today to find servers still running an affected release, then use a working exploit to confirm patches are effective and that endpoint and network controls would detect and stop an attack on any server that’s still exposed—connect with your Customer Success rep for help. Not a customer yet? Schedule a demo to see how the SafeBreach CTEM Platform turns breaking threats into measurable risk reduction. About the Author Noam Sagiv is a Product Manager at SafeBreach leading attack content for the SafeBreach CTEM Platform. Before SafeBreach, she spent five years in Israel’s Unit 8200, where she led teams building data-driven intelligence products used by more than 10,000 users. Frequently Asked Questions What is PHP2Shell? PHP2Shell is the name for CVE-2026-109618 and CVE-2026-109619, two critical memory-safety vulnerabilities in PHP’s core request handling disclosed on October 21, 2026. Both can let an unauthenticated attacker execute code on a vulnerable PHP web server with crafted HTTP requests. Researchers at Calif, OpenAI, and Wiz are credited with finding them. What does the CISA alert on PHP2Shell recommend? The Cybersecurity and Infrastructure Security Agency (CISA) alert “Critical Vulnerabilities in PHP Require Immediate Patching,” published October 21, 2026, directs organizations to update PHP to 8.5.12, 8.4.27, 8.3.36, or 8.2.35 immediately. For systems that can’t be updated right away, or that run PHP 8.1 or earlier, CISA advises taking them offline until they’re patched. It also recommends monitoring for PHP worker processes that launch shells or download tools. What are CVE-2026-109618 and CVE-2026-109619? CVE-2026-109618 and CVE-2026-109619 are memory-safety vulnerabilities in PHP’s core request handling, each rated 9.2 (Critical) under CVSS v4.0. CVE-2026-109618 abuses out-of-memory handling that leaves request memory uninitialized, and CVE-2026-109619 abuses global state that outlives the request memory it points to. Which systems are affected by PHP2Shell? CVE-2026-109618 and CVE-2026-109619 both affect PHP installations serving web requests, including PHP-FPM, Apache with mod_php, php-cgi, FrankenPHP, and LiteSpeed. The PHP project rates Apache with mod_php the highest risk. It has confirmed remote code execution on that setup, and exploitability on other setups depends on configuration. Is PHP2Shell being exploited in the wild? PHP2Shell exploitation reporting is still developing. Amazon Threat Intelligence reported exploitation attempts against its honeypots beginning October 17, 2026, four days before patches shipped, but said it was unclear whether those attempts would have succeeded. The PHP security team said proof-of-concept exploits published before the patches were not functional, and Dataminr reported no known compromises as of October 22. CrowdStrike has built a working PHP2Shell exploit in internal testing but has not released it. Has PHP2Shell been patched? Yes. The PHP project released fixes for PHP2Shell on October 21, 2026, in PHP 8.5.12, 8.4.27, 8.3.36, and 8.2.35. PHP 8.1 and older no longer receive updates from the PHP project, so organizations on those branches must obtain backported packages from a Linux distribution or long-term support vendor, build from PHP’s minimal source backports for 7.4, 8.0, and 8.1, or upgrade. PHP’s interim configuration changes reduce exposure but do not fully protect an unpatched system. How can organizations test their exposure to PHP2Shell? Start by confirming the PHP version actually serving each site, not the command-line version, and comparing it with the first fixed release for its branch. Then validate that controls detect exploitation and what follows it. Customers of the SafeBreach CTEM Platform can run two PHP2Shell simulations: one identifies web servers running an affected PHP release, and the other uses a working RCE exploit to test whether patches are effective and whether endpoint and network controls detect and stop exploitation. You Might Also Be Interested In Research Blog SafeBreach Coverage for #ProxyNotShell Vulnerabilities CVE-2022-41040 & CVE-2022-41082 Read More Solution Brief SafeBreach CTEM Platform Read More Research Blog SafeBreach Labs Releases Root Cause Analysis & Proof-of-Concept Exploit for CVE-2026-24061: Telnetd RCE as Root Vulnerability Read More Get the latest research and news

rcephpcvepoc @safebreach-blog 原文 ↗

无需认证、网络可达即利用,ASP.NET Core 部署面极广。

# CVE-2026-102870 - ASP.NET Core Remote Code Execution Vulnerability Improper control of generation of code ('code injection') in ASP.NET Core allows an unauthorized attacker to execute code over a network. CWE: CWE-94: Improper Control of Generation of Code ('Code Injection')

rcecve代码注入 @msrc 原文 ↗
必看2

压缩解码通常位于服务边界,畸形输入即可触发越界写。

# CVE-2026-111497 - Google: CVE-2026-111497 Buffer overflow vulnerability in the Brotli decoder Out-of-bounds write in .NET allows an unauthorized attacker to elevate privileges over a network.

越界写cvedotnet @msrc 原文 ↗
推荐2

通告覆盖从边缘设备入侵到横向移动的完整链路,可直接用于暴露面自查。

Oct 21, 2026 SafePay Ransomware: SafeBreach Coverage for CISA Alert AA26-293A Learn how SafeBreach maps CISA Advisory AA26-293A to simulations that test your exposure to SafePay ransomware’s edge-device exploitation, credential abuse, lateral movement, and data theft. Summary CISA Advisory AA26-293A, released October 20, 2026 by the FBI, CISA, the U.S. Secret Service (USSS), and German and UK law enforcement partners, details SafePay—active from August 2024 through 2026 and among the year’s most active ransomware groups. SafePay has impacted over 600 commercial enterprises and critical infrastructure organizations across North America, South America, Europe, and Australia. It is a high-complexity, multi-stage operation: edge-device CVE exploitation, credential brute-forcing, living-off-the-land binaries, and ransom. The goal is to exfiltrate sensitive files and extort payment. The impact is mass file encryption, destruction of local and NAS backups, data theft, and public leak-site extortion. This post covers SafePay’s four attack stages, its TTPs mapped to MITRE ATT&CK for Enterprise v20, the advisory’s IOCs, and the SafeBreach simulations—existing and newly added—that validate SafePay’s behavior from brute force through credential theft, lateral movement, exfiltration, and encryption. It also addresses two things the advisory implies but never states: why “we have MFA” does not answer this threat, and why a tested backup strategy does not close a double-extortion risk. Author: Tova Dvorin, Senior Product Marketing Manager On October 20, 2026, the FBI, CISA, the U.S. Secret Service (USSS), Germany’s Landeskriminalamt Rheinland-Pfalz, the UK’s National Crime Agency, and the UK’s North West Regional Organised Crime Unit released a joint Cybersecurity Advisory titled “ #StopRansomware: SafePay Ransomware ” (Alert Code: AA26-293A). SafePay is a double-extortion ransomware group that breaks in through VPNs, firewalls, and internet-exposed backup servers, then relies on legitimate admin tools to steal and encrypt data. The TTPs and IOCs come from FBI, USSS, and Landeskriminalamt Rheinland-Pfalz investigations as recent as August 2026—active law enforcement casework roughly two months old at publication, not indicators harvested from public reporting. Two of the six authoring bodies are investigative units rather than advisory writers. The advisory’s Key Actions are short: patch systems, especially firewalls and edge devices; implement Zero Trust; restrict administrative tools and privileges; and strengthen account security through ICAM policies and MFA. What is SafePay Ransomware? SafePay is a financially motivated ransomware group first observed in August 2024. It uses double extortion—stealing data, encrypting systems, then threatening to publish on its leak site—and has impacted over 600 organizations. It typically gains initial access through VPNs, firewalls, and internet-exposed backup servers. The group demands a subset of company revenue in cryptocurrency, leaves a ransom note on the desktop or root directory, and directs victims to a TOR .onion portal using a victim ID as the login. A representative makes phone calls to increase pressure. Non-payment means publication. What stands out in AA26-293A is how little of this needs custom tooling. PsExec, WinRM, AnyDesk, TeamViewer, ScreenConnect, RClone, WinRAR, 7-Zip, FileZilla, NetExec, AcronisDiskDirector—built-in utilities and commercial software, across nearly every stage. The authoring agencies call out this living-off-the-land (LOTL) approach explicitly, which means your detection surface is behavioral, not signature-based. They also caution anyone building blocklists from the IOC tables: these tools “should not be attributed as malicious without analytical evidence to support threat actor use and/or control.” The other defining trait is where SafePay starts. Every documented initial-access route touches the network edge—a VPN, a firewall, or an internet-exposed backup server. The 2026 Pattern: Three Advisories, One Front Door SafePay is not doing anything novel at the perimeter. It is reusing the most reliable entry point in enterprise security, and the overlap with other recent #StopRansomware advisories is specific enough to act on. Three of SafePay’s five CVEs also appear in the Akira advisory ( AA24-109A , published April 2024, updated November 13, 2025): Table 1: AA26-293A CVE overlap with the Akira advisory (AA24-109A) CVE Also exploited by Advisory CVE-2020-3259 Akira AA24-109A CVE-2023-20269 Akira AA24-109A CVE-2024-40766 Akira AA24-109A (Nov 2025 update) CVE-2023-27532 Not observed elsewhere AA26-293A only CVE-2024-40711 Not observed elsewhere AA26-293A only Akira’s advisory describes actors “obtaining initial access to organizations through a virtual private network (VPN) service without MFA configured, mostly using known Cisco product Common Vulnerabilities and Exposures (CVEs).” Its November 2025 update adds brute-forcing VPN endpoints. Gunra ( AA26-222A , August 10, 2026) makes three. Gunra exploited authentication bypass flaws in internet-facing firewall and VPN appliances, and Korea’s National Police Agency observed it “exploiting credential-exposure and SSH access-control weaknesses in internet-facing VPN gateways to obtain unauthorized remote access.” In one intrusion, actors reached an administrator account on an SSL-VPN appliance using default credentials, with no account lockout. Different crews. Different encryptors. Different countries of operation. Same front door. So a VPN and edge-device exposure review is not incident-driven work tied to one advisory—it is a standing control three separate ransomware operations have independently found profitable. If you patched for Akira in 2024 and 2025, three of SafePay’s five doors are already closed. If you did not, SafePay is the second group to walk through them. How does a SafePay ransomware attack work? A SafePay attack moves through four stages: initial access at the network edge, reconnaissance and privilege escalation, lateral movement and exfiltration, and destruction of recovery followed by encryption and extortion. Stage 1: Initial Access Attackers enter by brute-forcing exposed interfaces and running automated scripts against known edge vulnerabilities to bypass boundary authentication. The advisory adds that actors use Mullvad VPN IP addresses with phished account profiles to reach vulnerable edge devices. Stage 2: Reconnaissance and Privilege Escalation Actors map Active Directory network shares with ShareFinder.ps1 and SharpShares.exe, then extract administrator credentials via Mimikatz and DCSync to hold persistent control. Stage 3: Lateral Movement and Data Exfiltration The group moves laterally over RDP, PsExec, and WinRM, evading detection by packing executables and disabling host security software. Stage 4: Invalidation of Recovery and Extortion Actors dismantle recovery by terminating the Volume Shadow Copy Service and wiping NAS appliances to factory settings, then encrypt what remains, extorting victims via Tor and threatening public leak sites. SafePay TTPs AA26-293A maps SafePay to MITRE ATT&CK for Enterprise v20 . The techniques below come from the advisory’s Technical Details narrative and its ATT&CK tables (Tables 6 through 16). Initial Access External Remote Services ( T1133 ): Actors connect over VPNs, using Mullvad VPN IP addresses and phished account profiles to exploit known edge-device CVEs: CVE-2020-3259 (CWE-200: Exposure of Sensitive Information to an Unauthorized Actor) CVE-2023-20269 (CWE-288: Authentication Bypass Using an Alternate Path or Channel) Exploit Public-Facing Application ( T1190 ): Separately, actors exploit these across internet-exposed systems through automated scripts: CVE-2023-27532 (CWE-306: Missing Authentication for Critical Function) CVE-2024-40711 (CWE-502: Deserialization of Untrusted Data) CVE-2024-40766 (CWE-284: Improper Access Control) Brute Force ( T1110 ): Actors brute-force exposed interfaces and credentials. Note that while this functions as an initial-access route, AA26-293A’s ATT&CK tables map T1110 under Credential Access (Table 12). AA26-293A gives CWE classifications only and does not name affected vendors. Per NVD: CVE-2020-3259 and CVE-2023-20269 affect Cisco ASA and Firepower Threat Defense; CVE-2023-27532 and CVE-2024-40711 affect Veeam Backup and Replication; CVE-2024-40766 affects SonicWall SonicOS. All five are patched. The two routes look different in your logs. The credential-led route is human-driven and resembles a legitimate remote session. The scripted route is opportunistic and resembles internet background noise until it succeeds. Execution PowerShell ( T1059.001 ) and Windows Command Shell ( T1059.003 ): Actors run ShareFinder.ps1 to enumerate Active Directory file shares and identify exfiltration targets, and SharpShares.exe for the same purpose. System Services: Service Execution ( T1569.002 ): PsExec and similar tools execute commands on remote systems. Command and Scripting Interpreter ( T1059 ): Windows UAC elevates a program’s privileges to run malicious commands or binaries. Persistence Account Manipulation ( T1098 ): Actors gain administrative privileges on existing accounts. Create Account ( T1136 ): Actors create new accounts using built-in administrative tools, consistent with LOTL. ScreenConnect also supports persistence and access. Privilege Escalation Bypass User Account Control ( T1548.002 ): A UAC bypass creates an elevated process through a specified component object model, running commands or binaries at administrator level. Token Impersonation/Theft ( T1134.001 ): Actors have also used token impersonation. Stealth and Defense Impairment Disable or Modify Tools ( T1685 ): Before moving laterally, actors disable security software, modify or delete registry keys and configuration files, and disable security patches. Software Packing ( T1027.002 ): Packing or virtual machine protection conceals their code. System Binary Proxy Execution ( T1218 ): Execution is proxied through trusted, often Microsoft-signed binaries to bypass signature-based defenses. The advisory names PsExec and WinRM. File Deletion ( T1070.004 ): After encryption, SafePay may delete itself. Credential Access Brute Force ( T1110 ): Brute-force password attacks against accounts. Mimikatz harvests authentication credentials including Kerberos tickets ( T1550.002 ) . The IOC list includes dcsync.exe, “a tool used to perform DCSync operations to extract Active Directory credentials; high-confidence credential theft activity.” In some cases actors changed system user passwords, making valid logins impossible. Discovery Actors use SharpShares.exe, ShareFinder.ps1, netscan.exe, NetExec (nxc.exe), and nltest.exe to find hosts, shares, and domain trusts. The IOC list also includes their output artifacts—domains.txt, oui.txt, ping.txt, scan.xml, subs.txt—which are hunting targets in their own right. Lateral Movement Remote Desktop Protocol ( T1021.001 ): Actors use RDP with valid accounts and typically disable Windows Defender while using LOLBins. SMB/Windows Admin Shares ( T1021.002 ): SMB carries lateral movement through the network. Windows Remote Management ( T1021.006 ): WinRM executes commands on remote systems. Remote Services ( T1021 ): Remote services generally expand their presence. Collection Archive via Utility ( T1560.001 ): FileZilla, 7-Zip, RClone, and WinRAR compress and collect data. WinRAR splits data into segments and compresses to .RAR before exfiltration. Archive via Library ( T1560.002 ): Mapped by the advisory to credential dumpers such as Mimikatz. Exfiltration Over Alternative Protocol ( T1048 ): WinSCP and RClone move data to actor-controlled accounts; RClone syncs to cloud storage such as Mega. Actors have also run b2.exe to exfiltrate to Backblaze B2. Over Web Service ( T1567 ): OneDrive, in some instances. Actors install AnyDesk or TeamViewer on victim systems to transfer ransomware and other software in. Impact Inhibit System Recovery ( T1490 ): Actors halt the Volume Shadow Copy Service and reset NAS holding backup data to factory settings. The ATT&CK table goes further: in some cases they “deleted all network-attached backups of company data.” Data Encrypted for Impact ( T1486 ): A hybrid ChaCha20 and RSA scheme encrypts files, varying method by file type and size and supporting full or partial encryption. Files are appended with .SafePay and cannot be recovered without the key. What Third-Party Tools Does SafePay Abuse? Three tools carry most of the weight: Mimikatz: Harvested plaintext credentials, password hashes, and Kerberos tickets from memory to facilitate privilege escalation. PsExec: Executed remote commands and launched secondary payloads across compromised endpoints via Windows service execution. SharpShares: Conducted automated network host reconnaissance to discover accessible shares and map enterprise storage. The advisory’s Table 1 also lists 7-Zip, FileZilla, PowerShell, RClone, ScreenConnect, ShareFinder.ps1, WinRAR, WinRM, AnyDesk and TeamViewer, AcronisDiskDirector, NetExec, and Backblaze (b2.exe). SafePay IOCs IOCs observed between November 2024 and August 2026: Nine IP addresses —victim intrusions (104.26.2[.]223, 140.82.114[.]4, 185.199.108[.]133, 77.37.49[.]40), hosting victim data (173.194.194[.]94), client login panels (167.235.225[.]149), exfiltration (5.255.11[.]134, 37.10.71[.]188), and soc.dll command and control (88.119.167[.]239) Roughly 30 filenames , including locker.dll, dcsync.exe, Socks.exe, Socks_werfault.exe, sharefinder.ps1, netscan.exe, nxc.exe, b2.exe, DYRasWtb.exe, and netscanNEW.zip, plus legitimate binaries that are suspicious in context (CCleaner.exe, svchost.exe outside System32, nltest.exe outside admin workflows) Four SHA-1 and six SHA-256 hashes for the ransomware binary, locker.dll, soc.dll, SharpShares.exe, NetExec, a RunPE bundled DLL, and an injected PE Four actor hostnames: WIN-SBOE3CPNAL, WIN-3IUUOFVTQAR, WIN-DK18BI83M0S, WIN-I1HRRJHCEIV Download the AA26-293A STIX XML or STIX JSON from CISA for the full list. Two cautions govern how you use them. Verify IP addresses before blocking. And because malicious infrastructure often sits on cloud platforms with dynamically assigned addresses, “an address associated with malicious activity at one point in time may later be reassigned to benign services”—treat these as historically observed infrastructure, corroborated against current telemetry. This is the argument for detecting behavior rather than indicators alone. Addresses and hashes age in days. A crew’s preference for PsExec, RClone, ShareFinder, and WinRAR has held across two years of casework. “We Have MFA” Is Not an Answer to This Threat The advisory’s MFA guidance reads like a box already ticked. Most organizations running a VPN have MFA on it. SafePay still got into more than 600 of them. The gap shows in the mitigation language. The agencies recommend ICAM policies and MFA “for all services to the greatest extent possible,” then separately require all accounts with password logins—naming service accounts, admin accounts, and domain admin accounts—to meet NIST’s digital identity guidelines. They also recommend reviewing domain controllers, servers, workstations, and Active Directory “for new, unactive and/or unrecognized accounts.” Three recommendations, because enforcement coverage and policy coverage are different things. The accounts that break the pattern are familiar to anyone who has run an identity program: Break-glass accounts, deliberately exempted and rarely reviewed Service accounts authenticating to the same gateway as humans Local accounts on the appliance itself, outside the identity provider that enforces your policy Contractor accounts from a project that ended, still enabled Accounts an attacker already inside created or elevated—SafePay’s T1098 and T1136 activity means last quarter’s inventory is not today’s SafePay’s initial access compounds this. Phished account profiles over Mullvad VPN IPs defeat SMS and push-based MFA, which is why Akira’s guidance specifies phishing-resistant factors. And brute force only works where lockout thresholds are absent, which is why the advisory separately recommends failed-login lockouts. The question that produces an answer is not “do we require MFA on the VPN.” It is: enumerate every identity that can authenticate to an internet-facing service, and name the factor for each. The ones with no factor are your exposure. That list is usually short, surprising, and fixable in an afternoon. Backups Answer Encryption. They Don’t Answer Extortion. The reflex response to a ransomware advisory is to confirm the backup strategy. AA26-293A gives two reasons that is not enough. First, the backups are a target. SafePay halts the Volume Shadow Copy Service and wipes NAS holding backup data to factory settings; the ATT&CK table states that in some cases actors deleted all network-attached backups. Two of the five CVEs—CVE-2023-27532 and CVE-2024-40711—hit backup infrastructure directly. A backup server reachable from the internet is not a recovery control; it is an initial access vector holding your recovery data. Hence the advisory’s insistence on offline, encrypted, immutable backups in a physically separate, segmented location—and why “we back up nightly” and “we can recover” are different claims. Second, recovery does not undo disclosure. SafePay steals data before encrypting it. Restoring from a clean backup returns your systems. It does not un-publish the leak site, stop the pressure calls the advisory describes, or remove your notification obligations to regulators, customers, and employees. A company that restores perfectly and pays nothing has still suffered a data breach. The practical test: does your incident response plan have a path for “data stolen, systems still running”? Most have a restore runbook and an outage comms tree. Far fewer have a decision tree for the window between exfiltration and encryption, where detection is still possible and the outcome still changeable. That window is where the advisory’s exfiltration controls live: segmentation, filtering untrusted origins from remote services, and network monitoring that logs lateral movement. For boards, the reframe is simple. Backups are an availability control. Double extortion is a confidentiality event. Different investments answer them, and only one is usually funded. SafeBreach Coverage & Playbook Attack Updates Existing Behavioral Coverage SafeBreach provides simulations for much of this campaign. These are mapped to the ATT&CK techniques AA26-293A names and can be run now. Several, including #6513, #8857, #11773, and #7165, are Advanced Actions and should be scheduled accordingly. T1110 / T1021.001 —Brute force and RDP #192—Brute force attack over RDP protocol #11773—Brute force attack over RDP protocol against a live service (Windows) #6909—RDP Connection Between 2 Simulators #6473—Agentless lateral movement via RDP T1021.002 / T1569.002 / T1134.001 / T1550.002 —PsExec and SMB #1339—PsExec remote command execution traffic simulation #6513—Agentless lateral movement via SMB and RCE, using Mimikatz #6483—Agentless lateral movement via SMB and RCE, using Token duplication #6550—Agentless lateral movement via SMB and RCE, using Impersonated user #2273—Pass the Hash over SMB using Mimikatz T1021.006 —WinRM #5671—Agentless lateral movement via Windows Remote Management #5674—Agentless lateral movement via Remote PowerShell (WinRM) T1059.001 / T1059.003 —Share hunting and Mimikatz #11542—Network share discovery using net.exe (Windows) #1220—Inject Mimikatz using PowerShell to Extract Credentials T1098 / T1136 —Accounts #2189—Account Manipulation #2170—Create Local Account (Windows) #7170—Add a local administrator (Windows) #10794—Create Domain Account (Windows) T1685 —Disable or Modify Tools (Defender) #7834—Add Exclusions to Windows Defender #7835—Disable Windows Defender From Registry #8855—Add exclusion to Windows Defender via writing to registry #8854—Revert Windows Defender signatures #8857—Add Mimikatz as exclusion to Windows Defender and execute Mimikatz T1027 / T1070.004 / T1560.001 — Obfuscation, indicator removal, and staging #3829—Run obfuscated Mimikatz on host #2245—Indicator Removal on Host (Windows) #806—Collect Files using ZIP T1048 / T1567 —Exfiltration #103—Data asset exfiltration using FTP STOR (ASCII) #104—Data asset exfiltration using FTP STOR (Binary) #122—Data asset exfiltration over SSH #6975—Covert archived data asset exfiltration using HTTP/S POST #7200—Covert data asset exfiltration using HTTP/S POST file upload #8899—Sync a file outside of the OneDrive folder T1490 / T1486 : Recovery tampering and encryption #6372—Modify Volume Shadow Copy (VSS) #7165—Execute generic ransomware from a trusted path on user files Previously Covered IOCs One of the advisory’s hash-identified tools was already in the playbook before AA26-293A published. SharpShares.exe—the share enumeration tool from Stage 2—matches the advisory’s SHA-256 ending bc6f41: #10707—Write SharpShares (bc6f41) hacktool to disk #10708—Pre-execution phase of SharpShares (bc6f41) hacktool (Windows) #10709—Transfer of SharpShares (bc6f41) hacktool over HTTP/S #10710—Transfer of SharpShares (bc6f41) hacktool over HTTP/S #10711—Email SharpShares (bc6f41) hacktool as a compressed attachment #10712—Email SharpShares (bc6f41) hacktool as a compressed attachment Customers who have run these already have results against a named indicator. New IOC-Based Coverage We have added coverage for the SafePay locker payload, using the advisory’s locker.dll PE32 executable (SHA-256 ending f67917): #11836—Write locker_dll_ (f67917) ransomware to disk #11837—Transfer of locker_dll_ (f67917) ransomware over HTTP/S #11838—Transfer of locker_dll_ (f67917) ransomware over HTTP/S #11839—Email locker_dll_ (f67917) ransomware as a compressed attachment #11840—Email locker_dll_ (f67917) ransomware as a compressed attachment Together, the behavioral and IOC-based simulations cover SafePay’s reconnaissance, credential access, privilege escalation, defense evasion, lateral movement, staging, exfiltration, payload delivery, and impact. Coverage Scope These simulations validate SafePay’s brute force, credential theft, lateral movement, defense tampering, staging, exfiltration, and impact behaviors, including Volume Shadow Copy modification and real encryption of user files on the target simulator, along with the file-based indicators for SharpShares and the SafePay locker. How to Test Your Defenses Against SafePay Ransomware with SafeBreach SafeBreach customers can now validate their security controls against these TTPs in multiple ways. Method 1 – SafeBreach Scenarios: Navigate to the SafeBreach Scenarios page and choose CISA Alert AA26-293A. Method 2 – Attack Playbook: From the Attack Playbook, filter for the attack IDs listed in SafeBreach Coverage above. Method 3 — Known Threats Series: Open the Known Threats Series report, select the AA26-293A report, and run the simulations. Additional Advisory Steps Run the SafeBreach Platform Simulations Open the updated playbooks mapped to AA26-293A. Run simulations covering brute force, credential theft, lateral movement with PsExec and WinRM, Defender tampering, staging and exfiltration, and the locker payload IOCs. Start with #11836, #11773, #6513, #8857, #7200, and #7165 for coverage from payload delivery through encryption. Review results to validate detection, surface gaps, and guide remediation. Mitigation Strategies The authoring agencies recommend: Remediate known exploited vulnerabilities (KEVs) and keep systems, software, and firmware current, prioritizing internet-facing systems and the five CVEs in AA26-293A. CISA’s Binding Operational Directive 26-04 frames patch prioritization by technology type and forensics triage. Implement ICAM policies and require MFA across all services to the greatest extent possible, particularly webmail, VPNs, and accounts reaching critical systems. Hold all password-login accounts —service, admin, and domain admin included—to NIST’s digital identity guidelines: 15 to 64 characters, hashed in a recognized password manager, salted on shared credentials, never reused, with failed-login lockouts, no password hints, and no forced rotation more than annually. Maintain offline backups and test restoration regularly. Keep backup data encrypted and immutable across the entire data infrastructure. Run a recovery plan retaining multiple copies of sensitive data in a physically separate, segmented, secure location. Segment networks to control traffic between subnetworks and restrict lateral movement. Monitor for abnormal activity with a tool that logs and reports all network traffic, including lateral movement. EDR is particularly useful for lateral connections. Filter network traffic , blocking unknown or untrusted origins from internal remote services. Review domain controllers, servers, workstations, and Active Directory for new, unactive, or unrecognized accounts. Disable unused ports and consider an external-email banner. Audit administrative accounts and apply least privilege. Use time-based access for admin-level accounts and higher, such as Just-in-Time, supporting least privilege and the Zero Trust Maturity Model. Disable command-line and scripting permissions , since escalation and lateral movement depend on them. If You Detect a Compromise If compromise is detected before encryption, the agencies recommend: Isolate compromised hosts by quarantining or taking them offline. Threat hunt to scope the intrusion: artifacts, logs, TTPs, compromised devices and accounts, timeline. Report to the FBI and other agencies as appropriate. In the U.K., report to Report Fraud or a local police force. Apply eviction countermeasures with CISA’s Eviction Strategies Tool—Playbook-NG and the COUN7ER countermeasure database—once hunting data can inform the selection. Harden the network against further activity. If encryption has already happened, see the “Ransomware and Data Extortion Response Checklist” in CISA’s #StopRansomware Guide. The agencies do not encourage paying: payment guarantees nothing and may embolden adversaries to target others. Proactive Threat Monitoring Remote access tools: Flag unapproved AnyDesk, TeamViewer, and ScreenConnect installations, especially on servers. Staging and transfer tools: Alert on RClone, WinSCP, FileZilla, 7-Zip, WinRAR, and b2.exe, especially uploads to Mega, Backblaze B2, or OneDrive. Account changes: Investigate new accounts, privilege grants, and unexpected password resets on domain controllers. The advisory notes actors changed passwords to lock users out. Recovery tampering: Alert on Volume Shadow Copy shutdowns and changes to network-attached backup storage. Treat a NAS factory reset as a security incident, not a hardware event. VPN logins from anonymizing services: Review authentications from Mullvad and other commercial VPN ranges. Outbound transfers: Identify which hosts may open outbound SFTP, FTP, or HTTPS to arbitrary destinations. A file server making a transfer it has never made before is the clearest signal of exfiltration in progress. Stay Ahead with SafeBreach For a full view of your gaps against SafePay, sign into the SafeBreach platform and run the simulations mapped to AA26-293A. You can go further with SafeBreach Propagate , which enables you to assess how attackers could pivot across your environment post-compromise. It maps high-risk attack paths, visualizes lateral movement, and prioritizes remediation efforts based on exposure to your most critical assets. Learn more in the SafeBreach Propagate solution brief or schedule a demo . 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 CISA Advisory AA26-293A? A joint #StopRansomware advisory released October 20, 2026 by the FBI, CISA, the U.S. Secret Service, Germany’s Landeskriminalamt Rheinland-Pfalz, the UK’s National Crime Agency, and the UK’s North West Regional Organised Crime Unit. It details SafePay ransomware TTPs and IOCs from investigations as recent as August 2026. SafePay has impacted over 600 commercial enterprises and critical infrastructure organizations since August 2024. Which vulnerabilities does SafePay ransomware exploit? Five, all patched. CVE-2020-3259 and CVE-2023-20269 are exploited with phished account profiles over Mullvad VPN IP addresses; CVE-2023-27532, CVE-2024-40711, and CVE-2024-40766 are exploited across internet-exposed systems through automated scripts. The advisory lists CWE classifications rather than product names; per NVD, these affect Cisco ASA and Firepower Threat Defense, Veeam Backup and Replication, and SonicWall SonicOS. SafePay also brute-forces exposed interfaces. How does SafePay ransomware relate to Akira and Gunra? Three of SafePay’s five CVEs—CVE-2020-3259, CVE-2023-20269, and CVE-2024-40766—also appear in the Akira advisory, AA24-109A . The Gunra advisory, AA26-222A , describes authentication bypass flaws in internet-facing firewall and VPN appliances and credential-exposure weaknesses in VPN gateways. All three exploit the same class of entry point: internet-facing remote access with incomplete MFA enforcement and unpatched edge devices. Do backups protect against SafePay ransomware? They address encryption, not extortion. SafePay steals data before encrypting and publishes it if the victim does not pay, so restoring does not remove the disclosure or the notification obligations. SafePay also targets backups directly—halting the Volume Shadow Copy Service, wiping NAS to factory settings, and exploiting two CVEs in backup software. AA26-293A recommends offline, encrypted, immutable backups in a physically separate, segmented location. Which legitimate tools does SafePay ransomware abuse? Most central are Mimikatz for credential harvesting, PsExec for remote execution, and SharpShares for reconnaissance. AA26-293A also lists WinRM, AnyDesk, TeamViewer, ScreenConnect, 7-Zip, WinRAR, ShareFinder.ps1, netscan.exe, NetExec, nltest.exe, RClone, WinSCP, FileZilla, OneDrive, and Backblaze B2. The advisory cautions against attributing these as malicious without analytical evidence, so restrict and monitor them rather than blocking outright. How can organizations test their exposure to SafePay ransomware? The advisory’s Validate Security Controls section recommends testing your security program against its ATT&CK-mapped behaviors, continually and at scale in production. The SafeBreach CTEM Platform allows customers to select CISA Alert AA26-293A on the Scenarios page, filter the Attack Playbook by AA26-293A, or open the Known Threats Series report and run the simulations. Results show whether existing controls detect and block SafePay’s brute force, credential theft, lateral movement, exfiltration, and encryption behaviors. You Might Also Be Interested In Research Blog Gunra Ransomware: SafeBreach Coverage for CISA Alert AA26-222A Read More Solution Brief SafeBreach CTEM Platform Read More Research Blog SafeBreach Coverage for US CERT AA25-050A [Ghost (Cringe) Ransomware] Read More Get the latest research and news

勒索软件边界设备报告 @safebreach-blog 原文 ↗
更多2

Quote: Originally Posted by chefpopel27 Hey friends I'm trying to make a dma cheat runs quite well only my triggerbot is not so good I want him to have a vischeck so that he doesn't trigger you walls and which is very accurate on range. Can someone help me? Warm regards Check if the engine has a own check or bool for visibly and if not you gotta do it urself also can't you just get the pixel color of the crossbar and if its red shoot cuz im pretty sure thats the easiest method

游戏反作弊逆向 @unknowncheats-ac 原文 ↗