必看15
硬件钱包供应链遭物理植入,非官方渠道购入的设备可中招。
RT @BitcoinNewsCom: LEDGER HARDWARE WALLET IMPLANTS WERE DISCOVERED OVER A YEAR BEFORE THIS WEEK’S SUPPLY-CHAIN ATTACK REVELATIONS
Renowned hardware security researcher and computer engineer Joe Grand (@joegrand), also known as “Kingpin,” explains how he was contacted in August 2025 by a man in Thailand who had discovered a compromised Ledger Nano X.
The man had purchased a cheap Ledger online to replace the battery in his existing device.
When he opened it, he discovered a smaller battery, unauthorized circuitry, additional wiring, and an antenna.
After posting his discovery on Reddit and reportedly receiving little interest from local police, he contacted Grand to investigate.
The discovery came more than a year before the Ledger supply-chain attack became a major public controversy, with nearly $100 million in cryptocurrency reportedly stolen this week.
影响 Windows Server DNS 服务,缺官方补丁的版本可先微补丁缓解。
RT @0patch: Micropatches released for Windows DNS Server Remote Code Execution Vulnerability (CVE-2026-62878) https://t.co/U21oTKaBSf https://t.co/GEJ0n5avZv
大规模真实 Beacon 样本与配置,可直接用于检测规则和模型训练
Full details here: https://t.co/FU2tcQ536G
公开细节可复现,针对登录界面的本地凭据窃取新手法
Stealing Plaintext Windows Credentials
New Medium post, in this one we will see how to steal plaintext Windows credentials at the logon screen by abusing the Credential Provider COM interface and hooking internal methods inside CredProvDataModel.dll
https://t.co/fqPjueyhxF
凭据窃取windowspoc提权
@Salsa12__
原文 ↗
AI 自主挖洞产出完整链,攻防成本结构与自动化边界正在变化。
RT @gdb: 6 days and $40,000 of compute with GPT-5.6 Sol to find a critical vulnerability in Chrome, claiming Google’s first full-chain reward in 2026 for a vulnerability that would affect an estimated 1 billion users.
Highlights the opportunity for defenders to utilize these tools to secure their software to unprecedented levels:
ai_agent浏览器漏洞链
@moyix
原文 ↗
把 Agent 行为纳入检测视野,是本地 Agent 可观测性的现成起点。
So @Antonlovesdnb built an awesome project about telemetry (intent + action) for AI coding agents.
He also spoke about it in @alehurtadoxo's Detection Engineering Dispatch. You can find the full discussion in the reply. ⤵️
Project: https://t.co/OxWWYPe4GU https://t.co/qAO2F4MDRV
ai_agent工具检测遥测
@ipurple
原文 ↗
完整利用链的三个环节细节公开,可借其沙箱逃逸思路分析同类隔离边界
Very excited to share details on the bugs used in our full chain Chrome exploit! We used three bugs (V8 corruption, V8 sandbox escape, and renderer sandbox escape via Crashpad). The third bug is my favorite!
VBS 是 Windows 核心隔离基石,VTL1 层缺陷影响面广且细节首次公开
RT @Doyensec: 🔎 New Windows vulnerability discovered by Adrian Denkiewicz (@a_denkiewicz) of #doyensec - a high-severity flaw in one of Windows’ Virtualization-Based Security (VBS) enclaves that enables VTL1 code reuse. A first look at our ongoing research into this class of vulnerabilities.
🔗 https://t.co/l1cmjxc6o6
#appsec #security
windowsvbs逃逸
@_winterknife_
原文 ↗
全链赏金得主的完整打法与时间线公开,反映浏览器全链利用的现实门槛
RT @oegerikus: For the second time in Google Chrome’s history, Google awarded a $250,000 bounty for a full-chain exploit… to @XBOW. Here’s the full story:
At an all-hands at the start of 2026 in Malta, I set the team a crazy challenge. A full-chain exploit of Chrome by the annual Black Hat cybersecurity conference in August.
I believed there were few better tests of whether @XBOW could move beyond web applications to desktop software, servers and devices.
Chrome is one of the most hardened pieces of software on the planet, used by more than half the world’s population.
A full-chain exploit lets an attacker take total control of a machine, where they can read every email, password and bank login the user types. You don’t need to download anything or click a phishing link. Visiting the wrong site is enough.
So @moyix, Andres Luksenberg, Baul Lee, and Willem Pinckaers took on the challenge along with a fifth teammate: OpenAI's GPT-5.6 Sol.
They built a chain they thought would win, but Google told us they already knew one of the bugs.
We were already in Las Vegas for Black Hat and DEF CON, and the team didn’t want to go home without the win.
So Brendan and the agents built a new exploit in just 6 days.
We submitted it to Google and waited.
The news came through that we'd done what should’ve been impossible. A full-chain bounty of $250,000, the first awarded under Google’s 2026 rules.
It took just 6 days and $40,000 of compute to construct an exploit that affects a billion users of Chrome on Windows. An intelligence-agency veteran told me his team used to spend 6 months on this kind of vulnerability.
My takeaway: the cost of finding truly critical vulnerabilities is collapsing, for attackers and defenders alike. Restricting powerful models is not the answer; GPT-5.6 Sol is already public.
But defenders have one advantage attackers never will: they can test their own code first.
Assume your software is exploitable. Whoever is hunting your bugs has the same building blocks we did.
This is cause for urgency but not panic.
XBOW is now developing an autonomous product for testing compiled software (desktop applications, servers, embedded devices), and partnering with the world’s top organizations to design that product.
If you’d like to be one of them, get in touch.
P.S. Restart Chrome whenever the Update button appears!
同时给出植入搭建与蓝队检测思路,物理社工场景可复现
RT @g0h4n_0: New article on the blog about physical implants (dropbox) during Red Team engagements 🔥
📝 Physical Intrusion: A Laptop Charger Unlike Any Others!
How to build, configure and conceal a physical implant (dropbox) inside a laptop charger, Red Team side. And how to detect one, Blue Team side.
An old internal presentation turned into an article, with the Rust tool I use to get notified the second an implant comes to life: dende-rs 🦀
EN: https://t.co/Fb7vMxX7dy
FR: https://t.co/ukWbKMBLMe
dende-rs: https://t.co/utKXvNgOFw
红队物理渗透植入物工具
@0xocdsec
原文 ↗
用 MCP 与技能机制操控本机 AI 编码助手作 C2,载荷 URL 仍在线的
RT @Unit42_Intel: New #AI malware: C2Agent is an LLM-operated Windows malware with MCP, skills and living-off-the-AI, abusing the victim's own AI coding tools. Its payload URL, agent.x90[.]dev/data/copilot.txt, is still live as of this posting. More info at https://t.co/WHwin9JjNL https://t.co/R8FekoXUdv
ai_agentllm恶意软件
@artem_i_baranov
原文 ↗
满分 SSRF 且已遭在野利用,暴露的边界设备需立即排查
RT @securityaffairs: #SonicWall Patches CVSS 10.0 SSRF Flaw as Attacks Begin
https://t.co/FbFKFyBrUZ
#securityaffairs #hacking
@DefusedCyber @PrevidianCyber
ssrfcve边界设备
@DefusedCyber
原文 ↗
硬件钱包出厂即被植入后门的完整逆向案例,供应链攻击实证
RT @0xrootRE: A really excellent reverse-engineering report on a hardware backdoor implant in the LEDGER NANO X. @joegrand completed the reverse-engineering analysis of the backdoor implant in the crypto hardware wallet six months ago.
https://t.co/aApopqI5xF https://t.co/Pllm1xVnyO
Web 端/LibreOffice Online 常直接对外暴露,未升级即面临远程代码执行。
Remote Code Execution in LibreOffice!
Update to the latest version, especially if you're using LibreOffice in web applications. We found few and reported! https://t.co/IvKCNy9I1L
恶意加载器从 VS Code 任务延伸到开发机与 CI,构建即触发。
RT @marius_benthin: PolinRider has resurfaced in [email protected]. This version switches the disguised payload to fa-solid-300.llf and adds a second loader, api.js, to npm run build, extending execution beyond VS Code's folderOpen task to developer and CI builds. https://t.co/3O4V79GpXy https://t.co/uqxcsFyKYN
供应链npmai_agent
@cod3nym
原文 ↗
推荐15
从真实审计提炼的合约漏洞模式,可对照自查 v4 Hook 代码。
RT @jeffsecurity: Trail of Bits mapped 7 recurring Uniswap v4 hook bug patterns from real audits: missing caller checks, unvalidated PoolKeys, leaky delta accounting, logic in the wrong callback, mismatched permission bits.
https://t.co/KCcBD7F8FK
可下载的 C2 检测语料,适合做检测工程与机器学习训练
Dataset of 56,843 raw beacons and their parsed configurations 🤯
https://t.co/G09Eq5jgIM https://t.co/r0JRoMOme2
系统化免费课程,覆盖内核漏洞利用与固件攻防
RT @cr3ghost: If you're into Windows internals, reverse engineering or exploit development, bookmark this.
The @OpenSecTraining has an incredible collection of FREE low-level security courses covering:
- x86-64 assembly
- Windows kernel internals
- Windows kernel exploitation
- WinDbg from introductory to advanced
- IDA + Ghidra + Binary Ninja
- HyperDbg
- UEFI internals
- Intel firmware attack + defense
- C/C++ vulnerability research
- Symbolic analysis
- C++ binary reversing
- AFL++ fuzzing
- TPM + trusted computing
- Intel SPI analysis
- OS internals
- RISC-V
- Emulator development with QEMU
There is even a dedicated Windows kernel exploitation course covering race conditions + UAFs in KTM.
Proper low-level material, not another surface-level security course.
Start digging:
https://t.co/1zU3REgzUf
#WindowsInternals #ReverseEngineering #ExploitDevelopment
官方补丁清单与优先级方法,影响所有 Windows 资产维护
RT @msftsecresponse: Security updates for October 2026 are now available. Details are here: https://t.co/ue1BEvKRNp
As security update packages address more CVEs, prioritization can become more complex. In a new blog, we explore risk cohorts, an approach to grouping CVEs around shared deployment, exposure, and trust-boundary questions to help customers make vulnerability analysis more manageable.
Learn more: https://t.co/dsHO0kPev6
cve补丁微软报告
@artem_i_baranov
原文 ↗
从 Kerberoasting 到加固检测的可自建完整实验环境
A Purple Team lab simulating Active Directory attacks (Kerberoasting, BloodHound, lateral movement) and implementing defensive hardening (gMSA, tiering, LLMNR/NetBIOS disabling) with Wazuh SIEM detection rules mapped to MITRE ATT&CK. https://t.co/g5uKFTz9JK
免调试器解壳降低样本分析门槛,工具可直接上手。
vmp-lazy-unpack — Emulation-based unpacker and import resolver for Windows x86/x64 VMProtect 3.x binaries, using Unicorn to emulate syscalls and rebuild the IAT without a debugger. https://t.co/7Vv81SNTsz https://t.co/DQSZt2jbQh
展示 AI Agent 做漏洞利用的能力边界,对自建 Agent 做安全任务有直接参照。
RT @Jhaddix: Big thanks to our sponsor this week, @Xbow!
Really cool breakdown from the @Xbow team on finding a Linux kernel vulnerability, turning it into a working LPE exploit, learning about where autonomy still breaks down and how the team solved it.
https://t.co/SduWz5d41o
ai_agentlpe内核
@0xocdsec
原文 ↗
事件的一手技术细节,可看 AI 辅助漏洞利用的完整拆解。
RT @niemand_sec: Congrats to the whole team involved in this effort!!
Take a look at the blogspot here!
https://t.co/jl8AdK0xNu
ai_agent浏览器报告
@moyix
原文 ↗
AI 驱动浏览器利用的效率数据点,直接关系攻击成本下降。
Building a Google Chrome full chain would take a professional team around a year. At @Xbow we made it in 6 days
ai_agent浏览器漏洞利用事件
@fede_k
原文 ↗
把复杂派生哈希写成一行配方,简化自定义口令恢复建模。
RT @sekurlsa_pw: Hash recipes, hash-mode 4000 in @hashcat
https://t.co/4zxaA5DzGn
You can insert hash-recipes like (see examples and syntax in above link):
▪️md4(utf16le(pass))
▪️hex(xor(md5_bin(pass), md5_bin(salt)))
▪️md5(md5(salt) . pass) https://t.co/hMaihKNoFr
工具口令hashcat
@pentest_swissky
原文 ↗
检测逻辑 bug 有了开源定义与框架,可用于自建检测规则审计
🛡️ The ADE Framework provides a modern open-source formalization of Detection Logic Bugs - mismatches between what a detection rule intends to detect and what it actually detects.
https://t.co/GTOq9Uw5VD https://t.co/cAoqiYZ027
冷门注入原语,常规 API 监控易漏,适合免杀与检测规则研究
DLL Injection with SetWinEventHook: The Accessibility Hook as an Injector by @zodiacon https://t.co/bG2uhHdrar
windows注入规避
@ipurple
原文 ↗
官方长帖含漏洞组合与节奏,是浏览器全链利用的实操样本
RT @Xbow: How we chained three vulnerabilities into a full Chrome exploit in just 6 days—claiming Google’s first $250,000 full-chain reward of 2026 for an attack affecting an estimated 1 billion users. 🧵
漏洞情报 CLI 化,便于接入自动化流程做资产与 CVE 查询
cli — VulnCheck's official command line tool https://t.co/oFs92gkZCv https://t.co/m9hbHGE6QY
低成本铺开蜜罐获取全网攻击面数据,适合自建威胁观测
I got to catch up with @Andrew___Morris of @GreyNoiseIO to poke and play with their "Project Swarm": easy honeypots at scale and really neat network insights. Video here: https://t.co/qMYUT9Yqdm
蜜罐工具情报
@_JohnHammond
原文 ↗
更多61
AI 实验室高管会议室被物理窃听,暴露内部防护缺口。
abdulrahman alfozan, 35, is accused of bugging an openai exec room. a chatbot account tied to his wife asked about the recorder. after openai found it, they put in a camera. DOJ says he appears to be looking behind the tv here. all that prep just to get filmed. dumbass. https://t.co/yRSVQAi5GE
前沿 AI 实验室的内部威胁与网络安全短板被公开讨论。
idk about the other things jacob coxon says, but it’s funny seeing people call him delusional for suggesting there are spies in the labs and that labs are not secure enough in terms of cybersecurity. https://t.co/OAboylx9dp
开源反检测浏览器,用于爬虫与浏览器指纹绕过测试。
camoufox v156.0.1-beta.38 — 🦊 Anti-detect browser https://t.co/ttz3ue88xF https://t.co/s9RIhJ4iVp
可自托管部署的渗透测试环境工具。
Pwnbox is official finished, you can get it here!
If you have any recommendations you can contact me.
https://t.co/vYllvx4sIA
AI 聊天账号被用作作案工具,可作 LLM 滥用取证线索。
why does this read like a diy spy project? a chatbot account tied to his wife asked how to build and hide a recorder, then mentioned the MAC address of the one found at openai, they don't know who put him up to it. was he just bored? https://t.co/0qxSq0OyRo
llm事件内部威胁
@weezerOSINT
原文 ↗
把厂商能力宣称转为可落地的检测与响应投入方向。
tip: whenever anthropic publishes something that sounds delusional to many, price in half of it becoming reality within 6-12 months. then think about how you can be part of that world, copium and denial shouldnt be in our ally.
if you are wondering how i would price this in, i would probably dig more do some experiments to confirm their claims and start building incident response and threat monitoring teams/startups for the crazy world they talking about
it would be extremely funny if the guy didn’t have some grand ulterior motive to steal secrets.
the incentive could be as simple as running an anon twitter account where he leaks lab alpha and upcoming model releases before everyone else, just for attention.
RT @IntCyberDigest: ‼️ Pegasus maker NSO Group had access to the data Poland harvested with its spyware, PM Donald Tusk calls it "a crime against the state."
Experts found NSO kept sole control of the server that infects phones, and the data ran through its network abroad https://t.co/B0wffbpvxn
RT @ghostinthecable: Wanna know how my day is going? CrowdStrike just quarantined Defender.
AI 公司内部人员威胁实例,并用模型辅助设备取证
RT @OpenAINewsroom: We discovered a sophisticated recording device that was hidden inside one of our offices. We used our models to understand the device and where it had come from, identifying that a specific employee had planted it a few days earlier. We referred the matter to the FBI, and we’re grateful to the agents and prosecutors who moved quickly to make an arrest. The FBI’s investigation is ongoing and we will continue to support their work.
At this time, we have no indications of this individual inappropriately accessing or exfiltrating model weights, source code, or customer data. The employee was terminated.
We take protecting our technology, data, and people seriously, and as a matter of national importance. Our security program is designed to protect against a wide range of threat actors, from insider risks to cybercriminals to nation states, and this case is an example of how AI can help address these sophisticated security threats.
事件ai_agent内部威胁
@0xocdsec
原文 ↗
RT @JeffLadish: "this case is an example of how AI can help address these sophisticated security threats"
Spies are trying to steal our IP... also buy our products! 🙃
含注入链与 IOC,可直接用于检测和威胁狩猎
Full static & memory analysis of Overlord RAT:
✅Go agent architecture
✅HTTP/2 + WebSocket C2
✅Reflective DLL injection
✅Browser keylogging
✅IOCs
https://t.co/uuLm2R5c24
Oege's on TBPN now talking about the full chain! https://t.co/t2id1sQoTv
RT @BNONews: JUST IN: Saudi employee of OpenAI arrested after planting an audio recording device in the company's executive conference room - DOJ https://t.co/701PP7Pv99
RT @secbughunter: October Patch Tuesday is a big one: 1,400+ CVEs addressed.
That’s more CVEs in a single release than we shipped in all of 2025.
A huge thank you to the security researchers and Microsoft teams whose work made this possible.
https://t.co/cc3JYyyQEF
@msftsecresponse
RT @darkfloyd1014: Update, more bounty :)
https://t.co/mAVEWm1Y47
凭证失窃即直取私有仓库,暴露源码托管在第三方平台的现实风险面。
RT @IntCyberDigest: 🚨 BREAKING: Hackers broke into Meta’s private GitHub repo and stole their source code.
They got in with an employee's stolen credentials and took personal data on ~1,500 people, mostly US employees, stored there against Meta policy… https://t.co/ug8IRvaIeh
0-click 车机 RCE 的完整复现路径,车联网攻击面高价值参考。
RT @newbe3e: I've finally published the write up on my Pwn2Own exploit!!!
https://t.co/TjIydECFRs
(image unrelated) https://t.co/QnxXMjsfmc
影响三星设备图片解析,恶意图片即可触发内存破坏。
RT @ProjectZeroBugs: Samsung: https://t.co/FgEudFYExu Out-of-Bounds Write in Progressive JPEG Inverse-DCT from a Scan/Frame Component-Index Mismatch https://t.co/3qrua2KEPY
移动端内存破坏图片解析
@0xocdsec
原文 ↗
同库另一处内存破坏,构造 DNG 文件即可命中移动端解析链。
RT @ProjectZeroBugs: Samsung: https://t.co/FgEudFYExu Heap Buffer Overflow in DNG Uncompressed Tile Decoding (QuramDngReadImage::readUncompressed) https://t.co/7r1l1n28gN
移动端内存破坏图片解析
@FuzzySec
原文 ↗
语音助手提示注入的现成案例,助手可被引导越出既定立场。
RT @NatalkaKyiv: 🇷🇺 Alisa, Yandex’s AI assistant and the Russian equivalent of Alexa, doesn’t seem to be particularly loyal to Yandex… https://t.co/NJk48jMe8O
同一事件的媒体版本,补充 Black Hat 现场与厂商背景信息。
RT @Forbes: At the start of this year, Oege de Moor, CEO of $1 billion AI cyber startup Xbow, challenged his team to use AI to hack into the world’s most popular internet apps.
During the Black Hat hacker conference, one engineer used OpenAI’s Sol model to chain together three unpatched vulnerabilities in Google’s Chrome browser.
Read more: https://t.co/6FVlXPvGyx
Samuel Boivin/NurPhoto via Getty Images
ai_agent浏览器事件
@moyix
原文 ↗
no cyber benchmarking?
Researcher disclosures can be your advanced notice of viable threat models
may take a while longer because we are being bottlenecked by CI runners atm
have submitted a request for higher limits to Github
@S1r1u5_ @uz56764 That was the second one I know of, there should be one more
Wait who got the third full chain 👀 https://t.co/yQwRZZRfTg
shout out https://t.co/GBsUPrH6L9
Looking forward to this! Go read the post and start thinking up questions :)
i don't know for sure, but i am guessing anthropic was also using github?
its a good move to have their own source control, reduces the attack surface.
2023 年报告过的认证绕过 2025 年再现,指向真实在野的 ICS 攻击。
RT @Synacktiv: In 2023, @Synacktiv researchers identified an authentication bypass in #Danfoss SM800 and SC255/SC355 controllers. The same vulnerability, reported again in 2025, could be linked to the recent supermarket refrigeration attack.
🔎 Authentication Bypass - Danfoss SM800 & SC255/SC355: https://t.co/Gzf4DrntWp
🔎 Multiple Vulnerabilities - Danfoss SM800A: https://t.co/A3EsNHKiC4
绕过固件cveics
@0xocdsec
原文 ↗
含证书滥用与配置对比细节的恶意软件分析,可直接提取狩猎指标。
@malwrhunterteam shared a file signed "KATHRYN BANNER", revoked a few months ago. The signed file was a new build of #Spyder
We pivoted on indicators, pulled related payloads, pulled configurations, and compared them.
https://t.co/jHi69uTyj4 https://t.co/T5WYs6vHAE
恶意软件证书报告供应链
@SquiblydooBlog
原文 ↗
AI 助手可被社工绕过地图模糊处理,泄露敏感设施位置信息。
RT @nexta_tv: 🤖 Alice betrayed her own: Yandex's AI reportedly revealed the locations of the company's data centers to a supposed Ukrainian drone operator
On October 8, Yandex blurred the locations of its data centers on its maps, apparently in an attempt to protect them from Ukrainian strikes.
However, according to the independent outlet SOTA, a journalist posing as a Ukrainian drone operator asked Yandex's AI assistant, Alice, to help locate the company's remaining data centers.
The AI replied that it “understood the task,” provided the addresses of key facilities, and even wished the supposed operator good luck.
Alice, Alice, what have you done? 😂
ai_agentllm信息泄露
@0xocdsec
原文 ↗
图像地理定位能力商品化,公众住址可被单一照片反查。
RT @josephfcox: New: an AI company now lets cops identify specific houses, using data from Zillow/Airbnb, based on a single photo. It's part of Raven, previously called Geospy. Company previously demoed its tools to catch people vandalizing Flock cameras or protesters
https://t.co/ggbL2FdCYe
ai_agentosint隐私事件
@NetAskari
原文 ↗
可加入域名监控名单,提前拦截针对金融机构的钓鱼基础设施。
🚨 𝗙𝗿𝗲𝘀𝗵 𝗛𝗲𝗹𝗽𝗱𝗲𝘀𝗸-𝗧𝗵𝗲𝗺𝗲𝗱 𝗗𝗼𝗺𝗮𝗶𝗻𝘀 𝗜𝗺𝗽𝗲𝗿𝘀𝗼𝗻𝗮𝘁𝗶𝗻𝗴 𝗙𝗶𝗻𝗮𝗻𝗰𝗶𝗮𝗹 𝗜𝗻𝘀𝘁𝗶𝘁𝘂𝘁𝗶𝗼𝗻𝘀
While threat hunting with @silentpush , I came across an interesting cluster of 𝗻𝗲𝘄𝗹𝘆 𝗰𝗿𝗲𝗮𝘁𝗲𝗱 𝗵𝗲𝗹𝗽𝗱𝗲𝘀𝗸-𝘁𝗵𝗲𝗺𝗲𝗱 𝗱𝗼𝗺𝗮𝗶𝗻𝘀.
The domain names appear to reference major financial brands, including 𝗝𝗣𝗠𝗼𝗿𝗴𝗮𝗻, 𝗕𝗮𝗻𝗸 𝗼𝗳 𝗔𝗺𝗲𝗿𝗶𝗰𝗮’𝘀 𝗖𝗮𝘀𝗵𝗣𝗿𝗼, 𝗪𝗲𝘀𝘁𝗽𝗮𝗰, 𝗖𝗵𝗮𝘀𝗲, 𝗮𝗻𝗱 𝗛𝗦𝗕𝗖.
🛡️ 𝗧𝗼 𝗳𝗲𝗹𝗹𝗼𝘄 𝗱𝗲𝗳𝗲𝗻𝗱𝗲𝗿𝘀 𝘄𝗼𝗿𝗸𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝗼𝗿 𝗽𝗿𝗼𝘁𝗲𝗰𝘁𝗶𝗻𝗴 𝘁𝗵𝗲𝘀𝗲 𝗳𝗶𝗻𝗮𝗻𝗰𝗶𝗮𝗹 𝗶𝗻𝘀𝘁𝗶𝘁𝘂𝘁𝗶𝗼𝗻𝘀: this may be worth a closer look. Keep an eye on newly registered lookalike domains, helpdesk-themed lures, and related infrastructure appearing in your telemetry.
Sharing with the community for awareness and further investigation.🤝
#Threathunting 🧙♂️
浏览器崩溃上报组件被视为新攻击面,含 AI 参与漏洞研究的实践反馈。
@alkalinesec @fede_k Correct- that was one, mentioning crashpad as an attack surface was another, and giving it our first (non-rewarded) full chain to mine for exploit techniques was another. Also gave feedback on what bugs seemed worth pursuing
ai_agent浏览器攻击面
@moyix
原文 ↗
RT @lorenzofb: Great conversation here of the scoop I wrote last week.
If you have more information about GreyKey’s Preserve or how other companies get around inactivity reboot, my Signal is: +1 917 257 1382
https://t.co/vXG3xCfl3v
Based on the @Volexity writeup, we can now confirm genuine exploitation:
cve在野利用事件
@DefusedCyber
原文 ↗
覆盖注入、提权、TCC 绕过与持久化,macOS 攻击链工具再添一环
RT @BallisKit: Need to red team macOS targets but don't know where to start?
Introducing Mirage C2 🥷! Our new modular, in-memory macOS implant for DarwinOps.
Shell, SOCKS proxy, credential & secret extraction, plus private techniques for process injection, privilege escalation, TCC bypass & persistence.
From initial access to post-exploitation and EDR evasion, BallisKit now brings the full attack chain to macOS.
#RedTeam #macOS
macosc2工具逃逸
@BallisKit
原文 ↗
Running Tesla Model 3's Computer on My Desk Using Parts From Crashed Cars (@xdavidhu)
https://t.co/V308rKrc6s
#infosec https://t.co/q3zxcWrNXI
RT @DecipherSec: Look who it is! That @_JohnHammond guy from @HuntressLabs, talking to us about ClickFix attacks, AI-assisted attacks, and a future hacker movie episode!
https://t.co/gopCKSEz0U
clickfix钓鱼报告
@_JohnHammond
原文 ↗
9b model saturate exploitgym and exploitbench and get sota score, goo
then, i might change my mind on if it’s happening in next 2-3 months
llm基准ai_agent
@S1r1u5_
原文 ↗
there is a pattern in all the recent critical vulnerabilities coming out of gitlab.
look at the researcher(s) finding them and their track records.
and then ponder about the whole “yeah bro, just throw more tokens and you’ll saturate all vulnerabilities” argument.
gitlabllm观点
@S1r1u5_
原文 ↗
RT @_nnaci: Wrote up my attempt at deobfuscating Luraph-protected Lua: lifting Lua to LLVM IR , then using AI + partial evaluation to strip the VM, and checking it against the original.
https://t.co/sVwN6ihvxn
逆向混淆llm
@_winterknife_
原文 ↗
办公套件新 RCE,利用门槛低,等待补丁细节与 PoC
RT @b0n0b0__: Hey there @HacktronAI, I see that you too are having fun looking at LibreOffice! Me and my colleagues at @CodeanIO also played with that quite a bit BTW, here's a new RCE that we just reported, enjoy! https://t.co/ObsVhVM4lo
不新建线程的隐蔽执行手法,可用于更新检测规则
[Threat Intelligence - Fibers]
🚨 Windows Fibers can run arbitrary code inside an existing thread inside an existing thread, without creating a new one.
🕵️ Threat Actors | Earth Ammit & Earth Kurma
🌍 Country of Origin | 🇨🇳
🪲 Known Malware | CoffeeLoader & ACR Stealer
🎯 Sectors | Government & Technology
🗺️ Targeted Countries | 🇰🇷🇹🇼🇵🇭🇻🇳🇲🇾
windows绕过报告
@ipurple
原文 ↗
现成 Zoom 仿冒域名指标,可立即用于自查与封堵
⚠️ 𝗨𝘀𝗶𝗻𝗴 𝗭𝗼𝗼𝗺? 𝗕𝗲𝘄𝗮𝗿𝗲.
Defender XDR hunting has surfaced Zoom-themed domain indicators associated with 𝗦𝗮𝗽𝗽𝗵𝗶𝗿𝗲 𝗦𝗹𝗲𝗲𝘁 (𝟵𝟱), 𝗨𝗻𝗸𝗻𝗼𝘄𝗻 𝗮𝗰𝘁𝗼𝗿 (𝟮𝟱), 𝗮𝗻𝗱 𝗦𝘁𝗼𝗿𝗺-𝟮𝟳𝟴𝟳 (𝟮). Keep an eye on suspicious Zoom-related domains. 🫡
#threathunting 🧙♂️
出厂即被入侵动摇硬件钱包信任模型,涉及全部持机用户
Brace yourselves. This is going to be the biggest lawsuit in crypto history.
If @Ledger ships physical devices that arrive compromised at a customer’s door, they are finished. Period. Zero chance they walk away from this.
Even if you haven't been drained yet, every single user has grounds to sue for a blatant breach of trust and security.
用内置工具核查 Windows 权限,减少依赖不可信脚本
Just a friendly reminder: The SDDL returned by sc.exe is well-documented, with built-in self-explanation through the equally documented "showrights" parameter. Make good use of the built-in utilities. They’re far better than slop-coded magic tools. https://t.co/sGQGIt5b07
windows提权工具
@0gtweet
原文 ↗
RT @iss4c_f0ng: Successfully wrote a complete remote MS08-067 exploit from scratch with ROP...
calc.exe was spawned in session 0, so the window is invisible. I will write the blog post after I wake up...
#ROP #BoF #Exploit #MS08067 https://t.co/pwdHTLOw7L
poc漏洞利用windows
@J3rge
原文 ↗
REA is highlighting the incredibly transformative power of frontier models as well as the importance of open ones
what an exciting time
v6.5 will be out soon
brave-browser v1.99.36 — Brave browser for Android, iOS, Linux, macOS, Windows. https://t.co/BQcAr1wE0F https://t.co/rLRXrDuH1d
RT @poteto: i love writing. i never let agents write for me. whether it's docs, or tweets, or shitposts, or even prompts (okay maybe that one's harder as agents manage other agents)
there's a simple reason for that: i still strongly believe that writing is thinking. it's a skill i hope to never give up. it's also important to me that whatever i say is always authentic, mistakes and all. my favorite posts are the ones that i just full send without any editing
i suspect it also helps with prompting. being able to clearly express what i want, and knowing the vocabulary i need to be able to direct my agents does seem to make for more effective usage of agents. in the same way that engineers have an advantage when we prompt agents to write code, because we are intimately familiar with the words that make up the discipline
knowing the right word incantations to get an agent to produce a desired goal almost feels like spellcasting. as models get more capable, and that input box looks more and more like a magic box that can output anything, words become the interface between you and the genie
there may come a time when we give up the keyboard in exchange for brain-computer interfaces. but until then, i'm writing every word.
The owner bought the car used and asked for a basic cybersecurity check.
Nothing was wrong with the car mechanically. No warning lights, no battery drain, no strange behaviour.
But while going through the infotainment and telematics settings, I noticed something that would bother me if it were my own car.
The previous owner’s profile was still present.
That alone is not unusual.
People trade in cars every day without properly resetting the infotainment system.
What made this one more interesting was that the telematics module was still making outbound connections shortly after the ignition came on.
So I started looking at the traffic from the vehicle’s connected-services side.
The car was reaching cloud endpoints normally used for things like remote services, vehicle status, app syncing, and account-based features.
Again, that does not mean anything malicious was happening.
The real question was simpler:
Whose account was the car still associated with?
I checked the local infotainment profiles, paired phones, and connected-service settings.
There it was.
The previous owner’s account had never been removed.
That means the physical car had changed hands, but part of its digital identity had not.
Depending on the manufacturer and which services are enabled, connected-car accounts can be tied to things like vehicle location, remote lock status, maintenance data, trip information, charging information, or app-based controls.
Not every platform exposes the same features, but the principle is the same.
Ownership of the car and ownership of the digital account should match.
I documented the existing profile, removed the stale pairing, factory-reset the infotainment data, and verified that the telematics account was properly transferred.
After the reset, the old profile was gone and the expected onboarding process appeared instead.
This is one of those things people rarely think about when buying or selling a car.
They clear the glovebox.
They remove the toll tag.
They take out the garage remote.
But they forget that the car may also contain account data, paired phones, contact history, navigation destinations, Wi-Fi profiles, and cloud-linked services.
A used car should not only be mechanically inspected.
It should be digitally cleaned too.
Because sometimes the previous owner leaves more behind than a receipt in the centre console.
隐私车联网
@Officialwhyte22
原文 ↗
@_jensec I turned DeepSeek Hermes into a hacking tool 😅
Having fun (trying) reversing a ransomware this morning ;) Maybe its time to open the VM... https://t.co/2cuGvu5L0w
给出内核加固项被绕过的实测结论,可用于取舍实际部署的防护配置。
Linux Kernel Hardening Effectiveness from the Offensive-Defensive Perspective (NDSS paper)
https://t.co/gLyQ1Nd6Pl
#infosec https://t.co/xtFAszzkl3
自托管远程桌面常用库,版本跳变需核对本次修复项与兼容性。
FreeRDP v3.32.2 — FreeRDP is a free remote desktop protocol library and clients https://t.co/Kar71qYCz4 https://t.co/RuX4FjNdLo
点出 AI Agent 规模化攻防的执行底座这一走向,可作为选型线索。
Harness is absolutely required if you want to do hacking at scale.
ai_agentllm
@_jensec
原文 ↗
From one and only @sudhanshur705 and his agent swarms
ai_agent
@HacktronAI
原文 ↗
greptile configured and helping review the REA prs
a huge thank you to the Greptile team for reaching out so quickly and helping + @mvanhorn 🐐
@jihoonchoi @rahulbathijaa