⚡Megatron LLM Hub
All digests
Twitter

推特安全流

Top7

auth:api 路由的原生守卫被绕过,PoC 已放,存量 Laravel 站点可直接复现。

RT @HakaiOffsec: New Laravel one-day exploit by Yokai Research! 🔥 Full analysis, PoC, and exploit available here: https://t.co/MsaF6e2H3L We analyzed GHSA-f9x3-7vq9-cph6, a critical authentication bypass in Laravel affecting TokenGuard, the native mechanism used to protect auth:api routes. The issue was originally discovered by KIKOmanasijev(https://t.co/F0XhWDaiLr) and disclosed through GitHub Security Advisory. After the advisory went public, our team reproduced the vulnerable behavior, analyzed the authentication flow, identified the conditions required for exploitation, and developed a working PoC and exploit. We break down how the bypass happens, why TokenGuard can fail to enforce authentication as expected, and what that means for applications relying on affected configurations. No CVE has been assigned at the time of publication. Authors: @biscoitomesmo and @Gabriel52489026 Blog post: https://t.co/gdo3Z2ZvHb #HakaiSecurity #YokaiResearch #Laravel #SecurityResearch #authbypass #tokenguard

cvepoc绕过供应链 @0xocdsec source ↗

补丁未出而 PoC 在被复现,自托管 PHP 服务需监控子进程。

Since I heard about this #PHP #RCE last week, I've been preparing a tool for everyone who doesn't know what to do right now: process-watcher shows you every program your PHP processes start, with full command lines, so you can spot suspicious activity Until a patch is out, isolation (containers) and tight monitoring are your best bet to limit the damage. And I'm sure the LLMs already go "brrrr" and people are trying to reproduce the vulnerability. From what I've heard, exploitation will probably show up as suspicious sub process spawns. That's exactly what this watches, using eBPF: one static binary for Linux (amd64, arm64), no dependencies, running in a few minutes. It can also kill the ones you name (dry run first), but it's a detection layer, not a patch, and it can't see anything that happens inside PHP without starting a process. Quick start: https://t.co/7nluFV3Q25 Project: https://t.co/45agyRKtnS

phprce工具 @cyb3rops source ↗

对外gRPC服务默认暴露,写文件到RCE链路清晰,细节达PoC级。

We’re revisiting the #Fortinet #FortiSIEM for the fourth year in a row with a deep-dive on the recent advisory for CVE-2026-106173, an arbitrary file write vulnerability in a gRPC service, which leads to RCE. https://t.co/wTxRmKnJe4

rcecve固件报告 @Horizon3Attack source ↗

CVSS 9.2 认证绕过,默认配置受影响,修复版本 12.68.0/13.29.0

RT @kikomanasijev: I found a CRITICAL auth bypass in Laravel 🚨 (9.2/10) On MySQL/MariaDB, Laravel's token API guard let an unauthenticated attacker log in as an existing victim user. Patched in 12.68.0 & 13.29.0. Thanks to the @laravel team & @taylorotwell for the fast fix 🙏 More info 👇 https://t.co/tQ5ZL9Oz63

绕过cvephp @pentest_swissky source ↗

不调用注册表 API、无需管理员权限,EDR 看不到写入,红队原语同时是蓝队检测盲区。

RT @SEKTOR7net: Swarmer - persistence that never touches the registry. It turns a .reg export into an https://t.co/fDYv2QkQai hive that Windows loads over a user's own hive, abusing legacy Mandatory User Profile behavior. No registry APIs are called, so EDR sees no writes, and it works without admin rights. A post and tool by @praetorianlabs Source: https://t.co/cj0dT9L7v7 #redteam #blueteam

持久化绕过edr工具 @0xocdsec source ↗
Must-see24

用户态到 SYSTEM 的完整利用链与调优细节,可作 LPE 学习与检测样本。

RT @ale_sp_brazil: Exploiting Reversing (ER) Series | Article 11: Windows: Missing Bounds in the Print Spooler. The article has 60 pages, free. A user-mode target this time instead of a kernel one, worked end to end: from a missing bound check to a SYSTEM shell, with the service still running afterwards. Published on: https://t.co/VlebcByqKG What it covers: [+] The bug: a heap overflow (CWE-122) in the Print Spooler's form-enumeration path, reached through InternalEnumForms → CopyIniFormToFormInfo → PackStrings → memcpy. [+] It is still unpatched. I submitted it to Microsoft in March 2026 and MSRC assessed it as below the bar for immediate servicing, because the PoC requires administrative rights. Re-confirmed working in October against current mainstream and Canary builds. [+] The exploitation, which is two thirds of the article, with four stages: groom, corrupt, control, execute. The process never crashes: spoolsv.exe is left running and the planted form deleted. [+] Important note: I had already notified the MSRC months ago that I would write an article about the vulnerability and publish it. I would like to thank Ilfak Guilfanov ( @ilfak ) and Hex-Rays SA ( @HexRaysSA ) for the continuous support. This article will be published on https://t.co/GBDi0lO8Zd next week. Hope you enjoy the read! #Windows #PrintSpooler #VulnerabilityResearch #ExploitDevelopment #ReverseEngineering #HeapExploitation

lpe提权poc浏览器 @cyb3rops source ↗

厂商一年前拒收且至今可复现,PoC 与演示已公开,属内核提权面。

RT @AzizWho: 1/7 After months of research, I found a logic flaw in EasyAntiCheat: a standard user can control which older signed EAC driver its SYSTEM service loads into the kernel. Reported almost a year ago, rejected, and still reproducible. Demo: 2022 and legacy 2017 drivers loaded and running.

提权内核poc绕过 @0xocdsec source ↗

补丁发布后 EOL 实例仍会长期暴露,配套检测规则可提前布防。

My concern with the upcoming PHP RCE disclosure is how many systems could remain exposed long after patches become available. Roughly a third of PHP websites still run PHP 5 or 7. Some operators will need to migrate old applications before they can move to a supported release. Even on supported versions, plenty of systems won't be patched immediately. Some owners won't notice the advisory; others can't deploy an update that quickly. We still need the technical details to understand the actual exposure. If reliable, unauthenticated exploitation works across common deployments, I'd expect automated mass exploitation very quickly. AI-assisted exploit development and payload adaptation could shorten the response window further. For detection teams, I'd prepare now for the activity that follows. I expect plenty of familiar shells, downloaders, webshells and persistence mechanisms, even though the entry point is new. I've published a process-watcher guide with PHP configurations to record descendant processes and their command lines, with optional termination rules for shells, interpreters and known payload patterns: https://t.co/7nluFV3Q25 Short-lived commands can still be recorded, even when they finish before the kill response arrives. Activity entirely inside the PHP worker remains a blind spot. THOR Lite already includes relevant detections. We'll also review internally what additional post-exploitation coverage we can make publicly available. After patching, check for compromise. An update doesn't remove an existing webshell or invalidate stolen credentials.

rcephpcve报告 @cyb3rops source ↗

未授权即可打并给出 IoCs,公网 EBS 应优先排查。

We discovered CVE-2026-101767, an unauthenticated SQL injection in #Oracle E-Business Suite that chains to remote code execution. Fixed in Oracle's October CPU. Technical details to come later. Indicators of Compromise: 🔹 Unauthenticated POST requests to /OA_HTML/AppsTCFServer that instantiate the Gantt data-server class (oracle.apps.jtf.gantt.server.GanttDataServer), followed by backend database activity. 🔹 SQL reconnaissance shortly after unauthenticated web requests (database version, current user, directory objects). 🔹 Unexpected files in $APPLPTMP and the oracle account spawning /bin/sh via the Concurrent Manager to run OS commands.

sqlircecve @Horizon3Attack source ↗

详情已解禁,暴露面极广,PHP 8 以下版本确认无修复。

Okay, since the news is already out: an RCE vulnerability in PHP core, CVSS 10.0. Further details are scheduled for release on October 20. Only PHP 8 will receive patches. https://t.co/rluF5EpjGq

rcecvephp @cyb3rops source ↗

直接击穿 TLS 信任链,且自 2025 年签发以来长期未被发现。

RT @iangcarroll: It appears these intermediary certificates were issued with single-byte P-384 private keys. This means anyone can exploit this issue locally and issue valid certificates under them with no effort. This is very, very bad! Seems undetected since issuance in 2025?

tls证书供应链cve @cyb3rops source ↗

量化本地开源模型保留多少漏洞挖掘能力,直接降低自建成本。

RT @clearbluejar: I found two Windows 0-days with a frontier model, the first in 76 minutes. Then I replayed the same hunt on Qwen, DS4, GLM and more. How much frontier capability survives on local models, and how much human direction does it take? “To the Frontier and Back Again” @UnpromptedAU https://t.co/8m6u2F8BmV

llmai_agentpoc报告 @moyix source ↗

给出漏洞编号并确认可利用性,补丁发布前的排查窗口很短。

RT @watchtowrcyber: We were all supposed to wait until Thursday at 11am ET. Information has leaked wrt zero-day embargoed RCE vulns in PHP. We can confirm that they're real. CVE-2926-109618/619 We cannot share further information right now. Take this seriously and apply patches when available. https://t.co/kguM8fpq13

rcecvephp事件 @0xocdsec source ↗

结果附 PoC 与补丁,可直接用于自托管依赖的漏洞排查。

RT @ClaudeDevs: We launched OSS Scanner, which gives critical open source projects free, periodic vulnerability scans from our strongest models. Each finding includes a proof-of-concept, and a patch where possible. https://t.co/aonfpNG3WO

llm供应链工具poc @cyb3rops source ↗

沙箱逃逸直接打穿本地 Agent 隔离边界,细节已公开可复现

PwnEscape-Ai Research: Guest-to-Host KVM Escape (CVE-2026-99999). TLDR;- How a $1 million sandbox challenge led us to a vulnerability in Linux networking, repeatable host-memory disclosure, and ultimately a full guest-to-host escape. Blog:- https://t.co/CJERkG4rsE https://t.co/gdXbOvkklw

逃逸kvmcvepoc @5mukx source ↗

黑产工具已内置 MCP 服务端,账号接管与欺诈门槛进一步降低

Our latest research exposes how AntiDetect browsers enable fraud, account takeovers, and coordinated influence operations at huge scale. Knowledge bases, 24/7 support and MCP servers for LLM integration, the ecosystem is evolving fast. Read more --> https://t.co/g4vHtqwKx8

mcpai_agent浏览器报告 @_CPResearch_ source ↗

开源权重模型已具备真实攻击能力,本地部署方的威胁模型需重估。

RT @Hesamation: “We found DeepSeek V4.1 Flash 1017 beats GPT 6 Sol on our multi-step offensive cybersecurity task.” READ THAT AGAIN. the same cyber capabilities that OpenAI kept behind a gate got beaten by a Chinese open model you can run locally. buckle up guys. https://t.co/DDvax9fI8Y

llmai_agent红队 @cyb3rops source ↗

客户机到宿主 root 再到跨租户执行的完整链路,多租户云直接受影响。

RT @starlabs_sg: We built a full-chain QEMU/KVM escape on RHEL for Pwn2Own Berlin 2026. Guest to host root to cross-tenant code exec. ~100% reliable in 10s. Then came the toughest mitigation of all: the registration queue. 🤡 Our HITCON 2026 talk: “Pwn2OwNothing” 🧵

逃逸虚拟化云poc @cr0nym source ↗

进程认证模块的 UAF 可作提权原语,文章给出可迁移的利用构造思路。

RT @linkersec: When Defenses Become Attack Surface: CVE-2026-20971, a Samsung Kernel UAF Article about analyzing exploitation primitives derived from a racy use-after-free bug in the Samsung PROCA (Process Authenticator) kernel module. https://t.co/3SzgSG3G4k https://t.co/LbLCSGv37h

lpe内核cve移动端 @0xocdsec source ↗

完整暴露Agent编排、容器调度与单网关密钥构成的实战攻击架构。

Good report by @Ivanklydz on a Chinese actor's facilitating AI supported operations to go after government targets across Asia 'The platform ran 116 unique worker containers and logged roughly 230,000 dispatcher events over twelve days, consuming an estimated 36.7 billion LLM tokens through a single gateway key' : https://t.co/0sEVcFqT1U

ai_agentllm事件报告 @NetAskari source ↗

签名滥用叠加内核驱动落地链路完整,可直接转成检测规则。

PackClient RAT signed "Zaozhuang ZhongStore District HaoYouhao Department" Kernel driver signed "Xidao E-commerce Studio, Yishui County" https://t.co/k6scupTeTo Payload: 108.187.43[.]139:8080/china4[.]zip C2: 91[.]204.224.87:8790 156[.]251.16.214:8790 161[.]248.14.20:8790 https://t.co/ZolJGY7mWA

供应链提权内核事件 @SquiblydooBlog source ↗

自建邮件系统的常见前端,多个可组合的注入面且无 CVE 编号,补丁状态只能靠直查上游公告确认。

RT @Netlas_io: Multiple vulnerabilities in Roundcube, no CVE assigned 🤷🏻‍♂️ The Roundcube project today disclosed multiple vulnerabilities in Roundcube Webmail, including IMAP command injection, mail header injection, SSRF, CSRF, XSS, and other security issues. 👉 https://t.co/ukroXAi3gT

imapssrfxss自托管 @0xocdsec source ↗

把 AI 安全责任归于工程实现与厂商强制标准,影响后续工具与组件的合规叙事。

RT @derleomartin: JENSEN HUANG ZERLEGT MERZ Der Kanzler fragt den Nvidia-Chef auf offener Bühne, wie Europa KI regulieren soll. "Man kann nichts regulieren, wenn es die Technologie gar nicht gibt.“ Die Technologiebranche muss die Technik für Sicherheit selbst erfinden. Sicherheit besteht nicht aus Hoffnungen und Wünschen, Sicherheit ist Technik und Ingenieursarbeit. Zweitens brauchen die Unternehmen verbindliche Standards. Und erst zuletzt kommen Vorgaben von Industrie und Staat. Das Internet zu regulieren ist schwer, Internetanwendungen zu regulieren ist es nicht. Genau dort sollten wir ansetzen, bei den Anwendungen. Es gibt nichts zu regulieren, wenn die Branche nicht gedeiht. Man will doch, dass die deutsche KI-Industrie floriert, und zwar sicher und verantwortungsvoll. Wir wissen, dass wir Technik entwickeln und einsetzen können, die KI sicher macht. Wir wissen, dass wir uns als Unternehmen und als Branche an gute Standards halten können. Und wenn es dann noch nötig ist, regulieren wir." Merz stimmte zu und verwies darauf, dass Maschinen bereits aus dem AI Act herausgenommen wurden. Ein europäischer oder globaler Rahmen sei noch weit entfernt.

ai_agent标准监管 @0xocdsec source ↗
Recommended15

点明『验证签名者』与『该加载哪个驱动』的差别,附可复现 PoC。

RT @AzizWho: 7/7 This finding illustrates the distinction between verifying who signed a driver and verifying which driver a privileged service should load. I’m sharing the implementation and reproduction details for other researchers to examine. PoC: https://t.co/4Emvd0kwNO

提权内核poc绕过 @0xocdsec source ↗

开源模拟器做 Agent 沙箱的可行性进展,关乎本地隔离方案。

RT @momo5502: Can it run Crysis? Yes, Sogen can finally run crysis 🥳 I will probably write a blog post about what was done and what the future plans are. I feel like using Sogen as agent sandbox will soon become viable. And all that thanks for the help of countless contributors ❤️ https://t.co/fSH1oRKu4z

ai_agent沙箱工具 @daaximus source ↗

把 AI 对抗 EDR 的具体手法与 LLM 增益讲清楚,可直接借鉴。

RT @OffensiveAIcon: Your EDR Has Boundary Issues. @harmj0y and @tifkin_ of @SpecterOps explore how adversarial machine learning can evade EDR detection- and how LLMs are accelerating the reverse engineering behind it. Watch their Offensive AI Con talk below. 👇 https://t.co/cwnirhS595 #OAIC #OffensiveAI #Cybersecurity #EDR

llmedr绕过议题 @domchell source ↗

本地可部署的无限制模型带来真实漏洞挖掘增量,评测流程可复现。

RT @cybersec: DS4.1-1017 (abliterated) VulnHunter Results are IN!!! One NEW NOVEL / Unique vulnerability that was NOT discovered by a frontier model. TLDR; DS's trace files show it examined every one of these. The pattern isn't missed coverage. It saw the mechanism and dismissed the impact. Whenever I do a Cyber Eval there are two mechanims. The "easy" one is the benchmark on capability_matrix. The second is an operational pass to an oracle repo, to a pinned version, with known vulnerabilities. This is a Go repo for a production app. I used Grok as a "baseline" frontier model and all my open weight evals are measured against this.

llmai_agentpoc @cyb3rops source ↗

打包进二进制或模型的密钥不再是黑盒,自托管组件的混淆防护失效。

RT @attrc: "Security Through Obscurity Is Dead and LLMs Killed It." @moyix's @UnpromptedAU talk: a cat pic you can’t photocopy and an emulated HP printer that "runs out" of toner. Send this to your friends and coworkers who think their secrets are safe inside a compiled binary: https://t.co/oWoaVLfEHe

llm逆向ai_agent @moyix source ↗

开源 HTTP/3 内容发现工具,并附真实目标吞吐不达标的实测结论。

RT @assetnote: Our research team has released h3fuzz: a fork of ffuf with HTTP/3 support and a web UI for content discovery, based on PortSwigger's Turbo Intruder research. It reaches 120k requests per second on local benchmarks, and we cover why that rarely holds up against real targets. You can read more here: https://t.co/5gkPpDjkYv Repository: https://t.co/6M6KytDx7p

工具fuzzweb @0xocdsec source ↗

15分钟实机演示,具体说明LLM对逆向与漏洞发现门槛的削弱。

RT @caseyjohnellis: Woot! the @unpromptedau talks are starting to coming come online! 👀 This one by @moyix was probably my favorite: 15mins of unadulterated old-school “I’m hacking this because I think I can” followed by a one-shot live exploit demo involving a t-shirt with a cat on it worn by @mdowd. Seriously just go watch it (and subscribe for the others… More on the way!) Watch Brendan Dolan-Gavitt: Security Through Obscurity Is Dead and LLMs Killed It https://t.co/z5Nkp2qMAJ

llm逆向exploit议题 @moyix source ↗

补上攻击方Agent基础设施的配置失误,是可复现的防守切入点。

RT @Ivanklydz: We hacked the hackers infrastructure because they left their AI tool expose /tmp for exploit delivery. Read how: China-Linked Hackers Targeted Governments Across Asia https://t.co/CugcZGMzGu https://t.co/OJj4EccdV7

ai_agent容器事件 @NetAskari source ↗

少见的 AI Agent 实战化进攻案例,含行动规模、成功率与设施配置失误,可作为 Agent 攻防参照。

RT @vulonehq: A China-linked operator used AI agents to hit Vietnam’s Ministry of Defence, Taiwan’s legislature, and Indonesia’s intelligence agency. 66 separate operations. 48 got some level of access. Then the entire campaign was left sitting open on the internet. https://t.co/6WxzUz6HAp

ai_agent事件报告基础设施 @0xocdsec source ↗

给出 Agent 漏洞发现在各目标类型上的真实成败与误报来源,反编译质量是关键变量。

RT @AtipriyaBajaj: Earlier this year, we explored agentic vulnerability discovery across a diverse set of targets, such as Windows drivers, XNU, firmware, web apps, etc. We share our experiences here: https://t.co/bWsEEPLpA2 Fun fact: better decompilation can help reduce false positives!

ai_agent漏洞挖掘报告固件 @moyix source ↗
More82

把日志洪流收敛到关键事件的取舍方法,直接影响事件响应速度。

RT @andriinb: An enterprise environment emits billions of log events every day. During an incident, the responder rarely needs all of them. They need the few hundred or thousands that tell the story: the burst of 4625 failures followed by a single 4624 success, the encoded PowerShell spawned by winword.exe, the new service installed at 03:12 on a domain controller. Logs analysis is the discipline of getting from the first number to the second fast enough to matter. That job has become harder in two directions at once. Attackers using AI tooling generate evidence at machine speed and pad real actions with benign-looking activity. Meanwhile, the obvious shortcut of pushing logs into an LLM runs into guardrails, context-window limits and data-protection obligations. Hugging Face’s responders hit exactly that wall in July 2026, when roughly 17,000 , later to be claimed 3 billion events (still healthy daily workload) were refused by frontier models and had to be processed on a local model instead. Below, we cover log analysis end to end, then look at LogTotal, SOC Prime’s free answer to the incident-time version of the problem. https://t.co/wZaxcJwecZ

事件日志ir @cyb3rops source ↗

RT @johncrickett: My AI predictions. In the next 1 to 3 years, Open AI or Anthropic, or both, fail / get acquired for a fraction of their current valuations. Microsoft, Google or a developer tooling company with access to lots of developers and source code like Sonar or JetBrains wins AI in software engineering. Probably with smaller, local, specialised models. Other domain specific companies spring up to win domains such as mathematics, legal, writing, etc. TL/DR: The technology isn’t going away, but the financial bubble will burst. What do you think?

llm本地部署 @0xTriboulet source ↗

RT @doodlestein: Here's a project I've been quietly working on for months but finally got around to properly releasing: FrankenMermaid. It's a complete ground-up, memory-safe, hyper-optimized Rust implementation of MermaidJS that compiles into wasm. It's over 100x faster! https://t.co/EUc7AUl315 https://t.co/8MUjFx69h8

工具rustwasm @cyb3rops source ↗

商用模型对攻防内容收紧的具体案例,反衬本地自托管模型的可用性。

RT @_CryptoCat: Daybreak Blue just stopped working at the same time as all the other models stopped answering *anything* hacking related. When I go to @OpenAI cyber page it says my account is not eligible for individual access and to contact my org (old uni). Support are refusing any refund 💀

llmai_agent @0xocdsec source ↗

从用户态页到内核执行的完整手法,可照着复现。

Kernel Shellcode Execution via PTE Manipulation and HalDispatchTable New Medium post, today we are looking at how to promote a usermode page to supervisor executable via PTE manipulation, then redirect HalDispatchTable to execute a shellcode in ring-0 https://t.co/mvNhUge1WV https://t.co/iOs8u9Tzyh

内核lpepoc @Salsa12__ source ↗

驱动加载痕迹清理的具体步骤,属内核对抗常用手法。

🚀 Hiding your kernel driver from anticheat isn't as simple as manual mapping it. The next mandatory step to hide your tracks is clearing the PiDDBCacheTable. 👉 https://t.co/DXNInUGMTK https://t.co/cK1R0NsGos

内核绕过rootkit @GuidedHacking source ↗

自托管 LLM 推理栈又多一个可落地的开源选项。

RT @deepseek_ai: 🚀 Introducing DeepInfer, our in-house inference framework, now fully open source! Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable. 🔗 https://t.co/JGAik1kgk0 1/4

llm工具推理 @cr3ghost source ↗

RT @GoogleVRP: 📣📣📣 Calling all Chrome bug hunters! 📣📣📣 Rules updates: We're rewarding more types of deep security research, recognizing high-value fixes across more severity tiers, and providing new opportunities for researcher recognition. For details, see our latest blog post 👇 https://t.co/BwwD6gFDKz

浏览器赏金 @moyix source ↗

给出该技术的真实约束条件,比原始发布更利于评估威胁。

I spent some time yesterday testing the PhantomBuffer technique released by @TwoSevenOneT. A few observations: ✅ Requires elevated privileges ✅ ~32 KB implant size limitation ✅ Stealthy from a common API monitoring perspective ⚠️ Noisy from a debugger/telemetry perspective due to process attachment Interesting technique nonetheless. Full video below - Calc version:

edr绕过提权报告 @ipurple source ↗

不依赖框架的 BOF 实现路径,并解释它为何比 execute-assembly 更隐蔽。

RT @0xby7e_me: New post: Writing a Beacon Object File from scratch in C. COFF internals, dynamic import resolution, and building a privilege enumeration BOF - no frameworks, no wrappers. Plus why execute-assembly lights up every EDR while BOFs barely register. https://t.co/fuAJKeDaJ0

bof绕过edr工具 @0xocdsec source ↗

给出该样本的完整行为链,可作为检测与溯源参考。

RT @oneandonlyhusam: For anyone curious on how SilentXMR works I wrote a malware analysis report on it on openbin: https://t.co/NY5K8k5bQY It's a fascinating piece of goop.

恶意软件报告逆向 @0xpwnie source ↗

补充证书事件的根因细节,说明私钥空间仅 256 种取值。

RT @astuyve: Oh god: The p384 private keys are randomly selected integers between 1 and 256, the private keys are effectively all-zeroes apart from the last byte https://t.co/A1WXOLwugC

tls证书cve @0x64616e source ↗

提供 AI 产品安全设计缺陷的批评视角。

RT @robertgraham: The context behind such tweets is that the AI companies are rejecting 40 years of cybersecurity knowledge. They are ignorant of cybersecurity and are trying to re-invent it from scratch, repeating the mistakes of bad people from 40 years ago.

llm观点 @bohops source ↗

自建服务校验证书链时的基础依赖,便于核对根证书列表。

python-certifi — (Python Distribution) A carefully curated collection of Root Certificates for validating the trustworthiness of SSL certificates while verifying the identity of TLS hosts. https://t.co/m6pNJjnewd https://t.co/xYN8XOUmzA

tls工具证书 @KitPloit source ↗

HTTP/3 目标此前缺成熟发现工具,可直接替换现有 ffuf 流程

RT @infosec_au: You can now perform content discovery at 100k requests per second with h3fuzz on http/3 hosts (fork of ffuf with http/3 support). Tool here: https://t.co/tL4kkQRRwN Research post here: https://t.co/LSGVNco6G3 https://t.co/whSOcECvkO

工具fuzzhttp3 @pentest_swissky source ↗

把 IOC 溯源自动化到仓库分支与 PR 层,开源可直接接入狩猎流程

Skarvel, a tool developed by @twelvesec takes a single indicator of compromise, finds every public repository carrying it, digs through all their branches and pull requests, mines fresh IoCs from confirmed payloads, and keeps hunting https://t.co/IO4GIvOJRd

工具ioc威胁狩猎 @ipurple source ↗

现成 IOC 可直接做日志比对与威胁狩猎,覆盖面较广

🚨 𝗜𝗢𝗖 𝗔𝗹𝗲𝗿𝘁: 𝗖𝗟𝗢𝗨𝗗𝗛𝗔𝗭𝗘 (𝗤𝘂𝗮𝗿𝘁𝘇𝗟𝗶𝗻𝗲) ​ Sharing these @anyrun_app 𝗖𝗟𝗢𝗨𝗗𝗛𝗔𝗭𝗘 𝗜𝗢𝗖𝘀 for awareness and threat hunting. One highlighted sample shows prevalence across 𝟰,𝟱𝟭𝟱 𝘄𝗼𝗿𝗹𝗱𝘄𝗶𝗱𝗲 𝗱𝗲𝘃𝗶𝗰𝗲𝘀. ​ 🔍 𝗖𝗵𝗲𝗰𝗸 𝘆𝗼𝘂𝗿 𝗱𝗲𝘃𝗶𝗰𝗲 𝗻𝗲𝘁𝘄𝗼𝗿𝗸 𝘁𝗲𝗹𝗲𝗺𝗲𝘁𝗿𝘆 𝗮𝗴𝗮𝗶𝗻𝘀𝘁 𝘁𝗵𝗲𝘀𝗲 𝗜𝗢𝗖𝘀 𝗳𝗼𝗿 𝗽𝗼𝘁𝗲𝗻𝘁𝗶𝗮𝗹 𝗺𝗮𝘁𝗰𝗵𝗲𝘀. ​ #CLOUDHAZE #ThreatIntelligence #ThreatHunting

ioc威胁狩猎恶意软件 @0x534c source ↗

从误开的目录完整还原单人多阶段攻击链,TTP 可对照排查

RT @Huntio: 🇷🇺 One Russian-speaking operator. One rented server. Five months going after #Ukraine’s state data-exchange system We rebuilt the entire campaign from the directory they accidentally left open for 17 hours. 🧵1/4

威胁情报事件ttps @cyb3rops source ↗

嵌入式 TLS 库大版本改动经独立审计,固件选型与升级可参考

RT @quarkslab: Mbed TLS is a widely deployed open-source TLS, DTLS, and X.509 library targeting constrained and embedded environments. A few months ago, our cryptography team performed a security audit of the changes in version 4.0, sponsored by @OSTIFofficial ➡️https://t.co/xF0LYOFrK7

固件tls报告 @0xocdsec source ↗

变形重编译让签名检测失效,行为检测规则需要相应调整。

RT @objective_see: It’s not every day we encounter metamorphic ransomware targeting macOS! 👀 Our @patrickwardle recently presented at @macsysadmin & @virusbtn on “Heartbreaker,” a new ransomware sample that rewrites & recompiles itself every time it runs! Blog & slides: https://t.co/ghYQTrAXFL

恶意软件macos绕过 @patrickwardle source ↗

从悬垂读到泄露内核指针的回收手法,可迁移到同类 UAF 利用。

RT @eshard: Part 2 of our Android kernel exploit chain (CVE-2023-20938): the stale read returns garbage until you control what reclaims the freed slot. Land the right object there and it hands back a live kernel pointer instead. That’s your KASLR bypass. https://t.co/w3nxyKBsFI https://t.co/32jw0nCuQh

lpe内核cve移动端 @0xocdsec source ↗

针对主流 EDR 的规避工具已实战验证,检测侧需跟进其流量特征。

RT @Coiffeur0x90: I’ve been using ProxyBlob for a week on a target running CrowdStrike. The tool worked like a charm. Btw, it was made by my colleague @_atsika Definitely worth checking his tool 🚰 https://t.co/vfA0b1OpCI

绕过工具edr @cr0nym source ↗

自托管加密 DNS 代理更新,本地出口与解析链路可自行掌控。

RT @jedisct1: dnscrypt-proxy 2.1.19 released https://t.co/JYsKMcOP3E

工具自托管dns @cr0nym source ↗

移动端 TLS 分片实现简单,可作为流量对抗与检测研究的样本。

NoDPI4Android — Android proxy app that circumvents Deep Packet Inspection censorship by fragmenting TLS ClientHello packets and rewriting the TLS version to restore access to blocked HTTPS sites. https://t.co/cknIvBzeuC https://t.co/TgR1gsOihq

工具移动端绕过 @KitPloit source ↗

命名序列化与 IOC 明确,可用于钓鱼基础设施的早期拦截。

RT @Malwarehunterr: 🚨 Suspicious WhatsApp-themed domain cluster! Identified 473 domains registered in October 2026 and there are many more, reportedly hosted on 156.234.58[.]242 and redirecting to Yahoo or Bing search. The domains use sequential naming patterns and technical terms such as API, CDN, DNS, and security. Possible coordinated bulk registration, domain parking, or staged phishing infrastructure. Malicious activity remains unconfirmed. IP: 156.234.58[.]242 IOC: abkweb-whatsapp[.]com action-whatsapp[.]com aggregator-whatsapp[.]com ajz-whatsapp[.]com akb-whatsapp[.]com akc-whatsapp[.]com akd-whatsapp[.]com ake-whatsapp[.]com akf-whatsapp[.]com ala-whatsapp[.]com alb-whatsapp[.]com alc-whatsapp[.]com ald-whatsapp[.]com ale-whatsapp[.]com alf-whatsapp[.]com ama-whatsapp[.]com amb-whatsapp[.]com amc-whatsapp[.]com amd-whatsapp[.]com ame-whatsapp[.]com amf-whatsapp[.]com amg-whatsapp[.]com amh-whatsapp[.]com anc-whatsapp[.]com ane-whatsapp[.]com anf-whatsapp[.]com ang-whatsapp[.]com anh-whatsapp[.]com anj-whatsapp[.]com apibox-whatsapp[.]com apicloud-whatsapp[.]com apidev-whatsapp[.]com apiedge-whatsapp[.]com apihost-whatsapp[.]com apihub-whatsapp[.]com apilab-whatsapp[.]com apinet-whatsapp[.]com apinode-whatsapp[.]com apiops-whatsapp[.]com apisys-whatsapp[.]com appapi-whatsapp[.]com appcloud-whatsapp[.]com appdev-whatsapp[.]com appedge-whatsapp[.]com apphost-whatsapp[.]com appnet-whatsapp[.]com appnode-whatsapp[.]com appops-whatsapp[.]com appsec-whatsapp[.]com async-whatsapp[.]com audio-whatsapp[.]com authnode-whatsapp[.]com await-whatsapp[.]com bkz-whatsapp[.]com bla-whatsapp[.]com blb-whatsapp[.]com blc-whatsapp[.]com bld-whatsapp[.]com ble-whatsapp[.]com blf-whatsapp[.]com blg-whatsapp[.]com blh-whatsapp[.]com bli-whatsapp[.]com bliss-whatsapp[.]com blj-whatsapp[.]com blk-whatsapp[.]com bll-whatsapp[.]com blm-whatsapp[.]com bln-whatsapp[.]com blo-whatsapp[.]com blp-whatsapp[.]com blq-whatsapp[.]com blr-whatsapp[.]com bls-whatsapp[.]com blt-whatsapp[.]com blu-whatsapp[.]com blv-whatsapp[.]com blw-whatsapp[.]com blx-whatsapp[.]com bly-whatsapp[.]com blz-whatsapp[.]com bma-whatsapp[.]com bmb-whatsapp[.]com bmk-whatsapp[.]com bml-whatsapp[.]com bmm-whatsapp[.]com bmn-whatsapp[.]com bmo-whatsapp[.]com bmp-whatsapp[.]com bmq-whatsapp[.]com bmr-whatsapp[.]com bms-whatsapp[.]com bmt-whatsapp[.]com bmu-whatsapp[.]com bmv-whatsapp[.]com bmw-whatsapp[.]com bmx-whatsapp[.]com bmy-whatsapp[.]com bmz-whatsapp[.]com bna-whatsapp[.]com bnb-whatsapp[.]com bnc-whatsapp[.]com bnd-whatsapp[.]com bnf-whatsapp[.]com bng-whatsapp[.]com bnh-whatsapp[.]com bni-whatsapp[.]com bnj-whatsapp[.]com bnk-whatsapp[.]com bnl-whatsapp[.]com bnm-whatsapp[.]com bnn-whatsapp[.]com bno-whatsapp[.]com bnp-whatsapp[.]com bnq-whatsapp[.]com bnr-whatsapp[.]com bns-whatsapp[.]com bnt-whatsapp[.]com bnu-whatsapp[.]com bnv-whatsapp[.]com bnw-whatsapp[.]com bnx-whatsapp[.]com bny-whatsapp[.]com boc-whatsapp[.]com bod-whatsapp[.]com boe-whatsapp[.]com bof-whatsapp[.]com bog-whatsapp[.]com boh-whatsapp[.]com boi-whatsapp[.]com boj-whatsapp[.]com bok-whatsapp[.]com bol-whatsapp[.]com bom-whatsapp[.]com bon-whatsapp[.]com boo-whatsapp[.]com bop-whatsapp[.]com bor-whatsapp[.]com boxapp-whatsapp[.]com boxkit-whatsapp[.]com boxlab-whatsapp[.]com boxrun-whatsapp[.]com boxset-whatsapp[.]com boxuse-whatsapp[.]com branch-whatsapp[.]com bryweb-whatsapp[.]com btu-whatsapp[.]com btv-whatsapp[.]com btw-whatsapp[.]com btx-whatsapp[.]com bty-whatsapp[.]com btz-whatsapp[.]com bua-whatsapp[.]com bub-whatsapp[.]com buc-whatsapp[.]com bue-whatsapp[.]com build-whatsapp[.]com buz-whatsapp[.]com bva-whatsapp[.]com bvb-whatsapp[.]com bvc-whatsapp[.]com bvd-whatsapp[.]com bve-whatsapp[.]com bvf-whatsapp[.]com bvg-whatsapp[.]com bvi-whatsapp[.]com bvj-whatsapp[.]com bvk-whatsapp[.]com bwf-whatsapp[.]com bwg-whatsapp[.]com bwh-whatsapp[.]com bwi-whatsapp[.]com bwj-whatsapp[.]com bwk-whatsapp[.]com bwl-whatsapp[.]com bwm-whatsapp[.]com bwn-whatsapp[.]com bwo-whatsapp[.]com bwp-whatsapp[.]com bwq-whatsapp[.]com bxl-whatsapp[.]com bxm-whatsapp[.]com bxn-whatsapp[.]com bxo-whatsapp[.]com bxp-whatsapp[.]com cachesrv-whatsapp[.]com callback-whatsapp[.]com cdncloud-whatsapp[.]com cdnedge-whatsapp[.]com charm-whatsapp[.]com chathub-whatsapp[.]com clientapi-whatsapp[.]com clientcdn-whatsapp[.]com cloudapi-whatsapp[.]com cloudcdn-whatsapp[.]com clouddns-whatsapp[.]com cloudsec-whatsapp[.]com codehub-whatsapp[.]com column-whatsapp[.]com command-whatsapp[.]com commit-whatsapp[.]com cookieapi-whatsapp[.]com cookiecdn-whatsapp[.]com cookienode-whatsapp[.]com coral-whatsapp[.]com cors-whatsapp[.]com csrf-whatsapp[.]com cwjweb-whatsapp[.]com cycle-whatsapp[.]com dbcache-whatsapp[.]com decode-whatsapp[.]com devapi-whatsapp[.]com devcloud-whatsapp[.]com devedge-whatsapp[.]com devhost-whatsapp[.]com devnet-whatsapp[.]com devnode-whatsapp[.]com devpro-whatsapp[.]com devsec-whatsapp[.]com devsys-whatsapp[.]com dhcpnode-whatsapp[.]com discovery-whatsapp[.]com dmpweb-whatsapp[.]com dnscloud-whatsapp[.]com dnsedge-whatsapp[.]com dnshost-whatsapp[.]com dnsnode-whatsapp[.]com drift-whatsapp[.]com edgedns-whatsapp[.]com edgenode-whatsapp[.]com egrweb-whatsapp[.]com encode-whatsapp[.]com fhnweb-whatsapp[.]com flame-whatsapp[.]com format-whatsapp[.]com future-whatsapp[.]com giant-whatsapp[.]com gitnode-whatsapp[.]com gzlweb-whatsapp[.]com happy-whatsapp[.]com hostapi-whatsapp[.]com hostcdn-whatsapp[.]com hostly-whatsapp[.]com httpcdn-whatsapp[.]com httpsrv-whatsapp[.]com hubbox-whatsapp[.]com hubkit-whatsapp[.]com hublab-whatsapp[.]com hubpro-whatsapp[.]com hubrun-whatsapp[.]com hubset-whatsapp[.]com ids-whatsapp[.]com input-whatsapp[.]com ips-whatsapp[.]com ivory-whatsapp[.]com joker-whatsapp[.]com jqsweb-whatsapp[.]com kdyweb-whatsapp[.]com kitapi-whatsapp[.]com kitapp-whatsapp[.]com kitdev-whatsapp[.]com kitrun-whatsapp[.]com kitset-whatsapp[.]com kitweb-whatsapp[.]com knife-whatsapp[.]com labapi-whatsapp[.]com labbox-whatsapp[.]com labkit-whatsapp[.]com labrun-whatsapp[.]com labuse-whatsapp[.]com labweb-whatsapp[.]com latest-whatsapp[.]com layer-whatsapp[.]com layout-whatsapp[.]com legacy-whatsapp[.]com lemon-whatsapp[.]com level-whatsapp[.]com lexer-whatsapp[.]com limit-whatsapp[.]com linkhub-whatsapp[.]com lookup-whatsapp[.]com lpfweb-whatsapp[.]com macro-whatsapp[.]com magic-whatsapp[.]com mapper-whatsapp[.]com match-whatsapp[.]com merge-whatsapp[.]com mkwweb-whatsapp[.]com model-whatsapp[.]com mono-whatsapp[.]com nat-whatsapp[.]com netbox-whatsapp[.]com netlab-whatsapp[.]com netpro-whatsapp[.]com netset-whatsapp[.]com netsys-whatsapp[.]com netuse-whatsapp[.]com next-whatsapp[.]com nfdweb-whatsapp[.]com noble-whatsapp[.]com nodeapi-whatsapp[.]com nodecdn-whatsapp[.]com nodedns-whatsapp[.]com nodeedge-whatsapp[.]com nosqldb-whatsapp[.]com nqxweb-whatsapp[.]com okx-whatsaop[.]com opsbox-whatsapp[.]com opshub-whatsapp[.]com opskit-whatsapp[.]com opsnet-whatsapp[.]com opsnode-whatsapp[.]com opspro-whatsapp[.]com opssec-whatsapp[.]com opsset-whatsapp[.]com orchestrator-whatsapp[.]com output-whatsapp[.]com owyweb-whatsapp[.]com parse-whatsapp[.]com patch-whatsapp[.]com phase-whatsapp[.]com pipe-whatsapp[.]com podnode-whatsapp[.]com pool-whatsapp[.]com prev-whatsapp[.]com pride-whatsapp[.]com prohub-whatsapp[.]com prokit-whatsapp[.]com prolab-whatsapp[.]com promise-whatsapp[.]com prompt-whatsapp[.]com proset-whatsapp[.]com prouse-whatsapp[.]com proxyapi-whatsapp[.]com proxysrv-whatsapp[.]com pvnweb-whatsapp[.]com qtmweb-whatsapp[.]com quota-whatsapp[.]com range-whatsapp[.]com readme-whatsapp[.]com regex-whatsapp[.]com reload-whatsapp[.]com repoapi-whatsapp[.]com repocdn-whatsapp[.]com reponode-whatsapp[.]com resolver-whatsapp[.]com restart-whatsapp[.]com river-whatsapp[.]com rjxweb-whatsapp[.]com rule-whatsapp[.]com runapi-whatsapp[.]com runapp-whatsapp[.]com rundev-whatsapp[.]com runops-whatsapp[.]com runsec-whatsapp[.]com runuse-whatsapp[.]com sample-whatsapp[.]com schema-whatsapp[.]com scope-whatsapp[.]com sechub-whatsapp[.]com seclab-whatsapp[.]com secnet-whatsapp[.]com secnode-whatsapp[.]com secops-whatsapp[.]com secpro-whatsapp[.]com secsys-whatsapp[.]com serverapi-whatsapp[.]com servercdn-whatsapp[.]com sessionapi-whatsapp[.]com sessioncdn-whatsapp[.]com sessionnode-whatsapp[.]com setapp-whatsapp[.]com setdev-whatsapp[.]com setkit-whatsapp[.]com setrun-whatsapp[.]com setuse-whatsapp[.]com setweb-whatsapp[.]com signal-whatsapp[.]com smbweb-whatsapp[.]com softhub-whatsapp[.]com sort-whatsapp[.]com splitter-whatsapp[.]com sslcert-whatsapp[.]com ssltls-whatsapp[.]com stable-whatsapp[.]com stack-whatsapp[.]com state-whatsapp[.]com string-whatsapp[.]com style-whatsapp[.]com subject-whatsapp[.]com syncly-whatsapp[.]com syntax-whatsapp[.]com sysbox-whatsapp[.]com syshub-whatsapp[.]com syskit-whatsapp[.]com syslab-whatsapp[.]com sysnet-whatsapp[.]com syspro-whatsapp[.]com table-whatsapp[.]com tcp-whatsapp[.]com tcpip-whatsapp[.]com techly-whatsapp[.]com tgvweb-whatsapp[.]com theme-whatsapp[.]com tier-whatsapp[.]com tokenapi-whatsapp[.]com tokencdn-whatsapp[.]com tokennode-whatsapp[.]com topic-whatsapp[.]com tutorial-whatsapp[.]com uazweb-whatsapp[.]com udp-whatsapp[.]com udpip-whatsapp[.]com useapi-whatsapp[.]com useapp-whatsapp[.]com usedev-whatsapp[.]com userun-whatsapp[.]com usesec-whatsapp[.]com useweb-whatsapp[.]com vcnweb-whatsapp[.]com version-whatsapp[.]com vivid-whatsapp[.]com vpc-whatsapp[.]com vpcnode-whatsapp[.]com wafnode-whatsapp[.]com webact-whatsapp[.]com webadd-whatsapp[.]com webage-whatsapp[.]com webair-whatsapp[.]com webalert-whatsapp[.]com webant-whatsapp[.]com webape-whatsapp[.]com webarc-whatsapp[.]com webarm-whatsapp[.]com webart-whatsapp[.]com webash-whatsapp[.]com webask-whatsapp[.]com webbad-whatsapp[.]com webbag-whatsapp[.]com webban-whatsapp[.]com webbar-whatsapp[.]com webbat-whatsapp[.]com webcap-whatsapp[.]com webcar-whatsapp[.]com webcat-whatsapp[.]com webcompiler-whatsapp[.]com webdev-whatsapp[.]com webedge-whatsapp[.]com webengine-whatsapp[.]com webevent-whatsapp[.]com webframework-whatsapp[.]com webkotlin-whatsapp[.]com webmetric-whatsapp[.]com webops-whatsapp[.]com webparser-whatsapp[.]com webqueue-whatsapp[.]com webruntime-whatsapp[.]com webrust-whatsapp[.]com webscala-whatsapp[.]com webschema-whatsapp[.]com websec-whatsapp[.]com webstream-whatsapp[.]com webswift-whatsapp[.]com websys-whatsapp[.]com webtrace-whatsapp[.]com wheat-whatsapp[.]com xerox-whatsapp[.]com xrtweb-whatsapp[.]com xss-whatsapp[.]com ykbweb-whatsapp[.]com young-whatsapp[.]com zdhweb-whatsapp[.]com zebra-whatsapp[.]com #ThreatIntel #WhatsApp #Phishing #CyberSecurity @WhatsApp @500mk500 @skocherhan @KirkDerpca @0x534c

钓鱼ioc事件 @0x534c source ↗

RT @joshua_saxe: The big surprise of the 2022–26 era for AI security policy is how cyber damages to the public don't seem to have increased much due to AI. In fact, they've stayed on-trend with the industrialization-of-cybercrime growth rates we've observed since around 2015, with perhaps a small AI-causal bump. It's useful to think about the position of cybercriminals to develop an intuition for this: * Imagine you're trying to build a cybercrime business and you want to take advantage of AI automation. * How do you find and retain talented engineers who can build and operate the software/infra stack you'll need, given that they'll often have non-criminal alternatives with today's AI labor demand (even in middle-income countries where this demand also exists)? * Where are you going to rent space for your workforce to work? * Who are you going to hire who's willing to live there and live a life of crime? * If you're running your business remotely, how will you ensure honor among thieves and manage hiring and firing? * What's your strategy for evading law enforcement and bribing your way out of trouble? * How can you access AI inference in a stable way that allows you to innovate your work processes and tech stack? * Given all this, how will your quarterly automation goals compare to those of, say, a regional call center business in the American Midwest? There's a lot of nuance here (e.g., state and state-sponsored cyber attackers are a very different group; capabilities can get built up there and then unleashed discontinuously, and this needs to be considered). But overall, the demographic picture of cyber attackers means that folks should stop thinking of cyber attackers as a bunch of implicit MIT grads anxious to absorb the latest in automation to scale their businesses... (Below are some related charts on the demographics of cyberattackers from a blog post I'm working on.)

llm报告 @cyb3rops source ↗

RT @deepseek_ai: 🚀 Introducing DeepInfer, our in-house inference framework, now fully open source! Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable. 🔗 https://t.co/JGAik1kgk0 1/4

llm自托管工具 @cyb3rops source ↗

RT @justdionysus: I’ve seen arguments around VR and pay-to-win but I think defense (ie prodsec) is more pay-to-win than ever. The human bottleneck responding to the vulnpocalypse is dominating. It’s $$$ to triage (especially) and then fix a flood of issues in implementation and, worse, design.

报告供应链 @S1r1u5_ source ↗

覆盖进程注入隐藏与内存扫描规避,Windows攻防少见的系统化教材。

Proper old-school hacker energy. If you're into Windows internals, exploit development, malware memory or reverse engineering, Forrest Orr's (@_ForrestOrr) work is absolutely worth bookmarking. The FREE archive covers: - Windows memory corruption - Stack overflows + ROP - DEP, ASLR + CFG - Shellcode development - Windows internals - Malware memory artifacts - Phantom DLL Hollowing - Process injection + hollowing - Memory scanner evasion - Moneta memory analysis - PE internals - Browser exploitation - Chrome UAF exploitation - Firefox / IE exploit chains - Sandbox escapes - Defensive memory analysis And unlike a lot of modern security content, there are actual tools, PoCs, shellcode and exploit-development experiments to go with the research. Blog: https://t.co/cStzeIDi7t ExploitDev repo: https://t.co/JKaqBFO1QO Some of this is older material, but that's part of what makes it good. Proper fundamentals and low-level research that still teaches you how the machinery actually works. Huge credit to Forrest Orr (@_ForrestOrr) for keeping it publicly available. #ExploitDevelopment #WindowsInternals #ReverseEngineering

windowsexploit逆向工具 @cr3ghost source ↗

仿冒AI助手域名处于免杀窗口期,可作狩猎起点与员工提醒素材。

⚠️ 𝗜𝗻𝘁𝗲𝗿𝗲𝘀𝘁𝗶𝗻𝗴 𝗼𝗻𝗲 𝗳𝗼𝗿 𝗳𝗲𝗹𝗹𝗼𝘄 𝗱𝗲𝗳𝗲𝗻𝗱𝗲𝗿𝘀. ​ A domain registered just 𝟭𝟯 𝗱𝗮𝘆𝘀 𝗮𝗴𝗼, closely resembling a Microsoft Copilot FQDN. VirusTotal at capture time: 𝟬/𝟵𝟮 𝗱𝗲𝘁𝗲𝗰𝘁𝗶𝗼𝗻𝘀. ​ Sharing for awareness. 🛡️ ​ #ThreatHunting https://t.co/ubcUDZeUkx

钓鱼ai_agent事件 @0x534c source ↗

PDF、OLE、XOR解析与取证初筛一站覆盖,易嵌入既有分析流程。

DidierStevensSuite — Collection of Python scripts for malware analysis, file format parsing, hashing, and forensic triage, including PDF, OLE, and XOR analysis utilities. https://t.co/hbjKuedh2m https://t.co/j0n4gTT6Ds

工具恶意样本取证 @KitPloit source ↗

展示自托管Agent自主逆向API并编写客户端的实际能力边界。

RT @sighmon: It works! @bot added my B-hyve XD HT31 Bluetooth watering system to Apple HomeKit using @brutella ‘s hap. I showed it @morluto ‘s rea, it reversed the official Orbit app, wrote a key requester using my account details and the official API, and now happily runs the Go app on my Raspberry Pi.

ai_agent逆向移动端 @morluto source ↗

鉴权、输入校验、框架默认值逐条验证,低成本高回报的检查项。

Assumptions are at the root of so many security issues. When testing an application or reviewing code, always ask yourself: "What assumptions did the developer make? What assumptions am I making when I say this is secure?" Maybe you're assuming that an input has already been validated, that a route requires authentication, or that a framework handles a security check for you. Each of those assumptions deserves to be verified. Questioning the developer's assumptions is part of the job, but questioning your own is just as important.

方法论代码审计 @PentesterLab source ↗

给出对抗Agent化攻击的防守优先级排序,可作策略参考。

RT @ZackKorman: There is only one answer to cybersecurity threats: Fix your shit. Scared of threat actors using AI? Worried about rogue agents? Think abliterated models are spooky? Fix. Your. Shit. The "ban AI models and regulate GPU access" crowd should instead fix their shit.

ai_agent方法论 @0xTriboulet source ↗

含 1187 次尝试、10 次成功利用与 9 起窃取的后端日志分析

The SlowMist team linked DarkSword's panel to public GitHub code. 1,187 attempts, 10 successful exploits and 9 thefts were logged. https://t.co/DNcSpX1hnK https://t.co/GZb9xg9phm

移动端恶意软件报告ios @virusbtn source ↗

模块化凭据窃取家族,附 C2 域名可直接用于检测

RT @abuse_ch: Nyxar Stealer with potential Turkish roots 🇹🇷, sold as malware-as-a-service with of 60 modules, including: 🪝 Browser credential and cookies stealer 💬 Messaging app credential stealer (Discord, Telegram, etc) 👛 Crypto wallet stealer 🎮 Gaming app credential stealer (Steam, Epic, etc) Probably of turkish origin: 🕑 Client quotas reset at 03:00 Instanbul time 🇹🇷 tr-TR formatting and error messages in bot admin panel Botnet C2s: 📡nyxar .cc 📡nyxar .lol 📡nyxarstealer .lol 📡nyxarst .lol 📡polnet .lol Malware samples: https://t.co/89iA4Y8kQR

恶意软件凭据窃取c2 @virusbtn source ↗

RT @calif_io: Calif was founded with a mission to make the Internet a more secure place for everyone. In the last few months, we have been seeing significant advances in AI model capabilities, especially in the domain of offensive security research. In addition to publishing our research and learnings, we wondered what else we could do to help defenders. While large corporations and government agencies have the dedicated budget and human resources for incident response and forensics, there are non-profit organizations –- collectively known as civil society organizations –- investigating digital threats affecting vulnerable groups and they are funded entirely by volunteers. In order to perform their job, the civil society researchers may need to extract data from mobile devices for forensic purposes, reverse engineer malware, and conduct similar security tasks. This work can be challenging given the restricted environments on some mobile devices. We understand these technical challenges and will offer our research services and capabilities to these organizations, free of charge. We support them and their mission. If you represent such an organization and need assistance, contact us at [email protected].

llm报告 @0xocdsec source ↗

零耦合模块设计,可拆出浏览器指纹与硬件特征采集组件用于红队钓鱼页或隐私测试。

SilentSniffer — A professional, high-fidelity de-anonymization and vulnerability diagnostic framework. Features a zero-coupling modular architecture for browser-based intelligence gathering, hardware fingerprinting, and security research. https://t.co/cy1g6Dc669 https://t.co/7mnebYDQmg

工具指纹浏览器反匿名化 @KitPloit source ↗

关于商用大模型在应急响应中可用性的争议观点,缺细节,只能当作选型风险提示。

RT @0sm0s1z: I would NEVER use an @AnthropicAI or @OpenAI model for security purposes. If you’re relying on this (daybreak or no daybreak) you have a critical point of failure in your enterprise IR. This has been proven in multiple breach cases now. Ditch them.

llmir供应链 @0xocdsec source ↗

受影响面窄但对以 OpenBSD 为边界设备的环境是例行升级节点,需看安全相关变更。

RT @openbsd: OpenBSD 8.0 is out now! See what's new: https://t.co/Zjy2TdgZbd https://t.co/1IwZ74VuRx

openbsd发布 @0xocdsec source ↗

RT @nvidia: Germany knows how to build, and now it’s bringing that expertise to industrial AI. 🇩🇪 Tonight in Berlin, @JensenHuang and Chancellor @_FriedrichMerz discussed combining German engineering with NVIDIA AI to advance manufacturing and automotive. We’re also backing Germany’s builders, with new planned investments in Agile Robots SE, @aveliosmed, @MireloAI, and @reflexaerospace, spanning robotics, healthcare, creative tools, and space. Meet the companies and catch up on today’s news from Berlin: https://t.co/hTnsNMbPa6

llm产业动态 @0xocdsec source ↗