elfe @elfe2000

Since some people assumed this is a 0day in V8, Im gonna clarify that it isnt one. Deepseek chained known vulns into a working exploit. Thats it :)

Exocija @Exocija

5 days ago

I told Deepseek V4.1 Flash 1017 to build a exploit that escapes Chrome's V8 Sandbox After 12 hours, it had a multi stage memory corruption exploit that achieves full process read and write.

48 82 2K 130K 747

I can exploit this with full weaponization. @LogitechG hi! 👋

Deniz M. @dennywise0

5 days ago

Now this turns into a LPE on a fully patched fresh Windows installation. @LogitechG love ya (I swear I’m not cheating the UAC bypass look at my hand😭) #zeroday #CyberSecurity #infosec

0 0 20 2K 3

Proof of Concept for CVE-2026-21509 vulnerability, bypassing Windows Defender PoC: github.com/gavz/CVE-2026-…

I told Deepseek V4.1 Flash 1017 to build a exploit that escapes Chrome's V8 Sandbox After 12 hours, it had a multi stage memory corruption exploit that achieves full process read and write.

A PoC/exploit has been discovered for vulnerability CVE-2026-98260 PT ID: PT-2026-106589 Read on dbugs: dbu.gs/vulnerability/… Vendor: Linux Product: Linux kernel Description: A use-after-free issue exists in the Linux kernel during the execution of execve(). When a non-leader thread performs an exec(), the de thread() function changes the task that owns the Process ID (PID). If subsequent operations such as unshare files(), set mm exe file(), exec mmap(), or exec task namespaces() fail, the task may enter do exit() before the POSIX timers are cleaned up. This leads to a scenario where exit itimers() frees the k itimer while its node remains queued. When the task is later reaped, cleanup timerqueue() calls timerqueue del() on the already freed node, resulting in a use-after-free condition. Link: github.com/V4bel/Infinity…

CVE-2026-81376: VS Code RCE via Security Feature Bypass; $4,000 bounty. Write-up: samchen.blog/blog/vscode-ne…

This is why the anti-cheat arms race is getting so interesting. Riot is pushing Vanguard further into hardware-backed trust, attestation, VBS/HVCI, IOMMU and deeper platform security. At the same time, AI-assisted reverse engineering is making it much faster to tear apart the software sitting on top of those protections. @k0mkc is showing heavily devirtualized VGK internals and says he can now devirtualize the entire thing. On the EAC side, @in19h has published tooling that recovers large parts of a heavily virtualized Easy Anti-Cheat Linux sample into readable code and IDA artifacts: github.com/19h/eac-linux-… There is an important parallel here with EDR. An EDR is only as effective as the detection engineering, threat hunting and R&D behind it. Anti-cheat works the same way. You need researchers constantly following new techniques, detection engineers turning that research into useful telemetry and detections, and people actively hunting for behaviour that existing rules may miss. That means continuously keeping up with: - new cheat architectures - kernel abuse - BYOVD - hypervisors - DMA - firmware tricks - input manipulation - virtualization + obfuscation - new bypass techniques The game was already there. AI is speeding it up and lowering the barrier to reversing, classification, devirtualization and research. That forces both sides to iterate faster. A lot of respect to the Vanguard team as well. They have clearly been keeping up with some of the most cutting-edge work in Windows security, hardware-backed trust, kernel security and anti-cheat research, and their recent Vanguard On-Demand changes show how quickly the defensive side is evolving too. This is becoming a much faster research cycle on both sides. Where does the arms race go next? Huge credit to @k0mkc and @in19h for the research. #AntiCheat #ReverseEngineering #WindowsInternals

keyboard_arrow_left Previous keyboard_arrow_right Next

cr3ghost @cr3ghost

4 months ago

BREAKING NEWS: Riot just announced Vanguard On-Demand. Vanguard no longer needs to start at boot. If you enable Secure Boot, TPM 2.0, VBS, HVCI, and IOMMU, Vanguard will only run while you play. Microsoft built a Runtime Driver Attestation Report that lets Vanguard verify driver

keyboard_arrow_left Previous keyboard_arrow_right Next

8 63 410 83K 247

Ai just found a full Chrome exploit chain that affected a billion users. - It took 6 days and $40k. - Google paid us $250k for it. Three different bugs. All of them involved mutating something the browser assumed was immutable. Found by : xbow.com/labs/research/…

XBOW @Xbow

2 weeks ago

ICYMI: We found a full chain Google Chrome exploit in just 6 days that affected a billion users. Here are 4 things we learned. 🧵

2 19 205 26K 84

NTLM Reflection Attack 🔥 Telegram: t.me/hackinarticles ✴ Twitter: x.com/hackinarticles NTLM Reflection is a technique where attackers trick a system into authenticating against itself, allowing privilege escalation without knowing user credentials. ⚡ Attack Highlights 🎯 Coerce victim machine to authenticate 🔄 Reflect NTLM challenge back to same system 🎟 Reuse authentication response 🔐 Gain authenticated session as victim 🚀 Escalate privileges to SYSTEM 💡 Reflection attacks exploit flaws in challenge-response authentication, where a system unknowingly validates its own authentication request. 📖 Article: hackingarticles.in/ntlm-reflectio… #CyberSecurity #ActiveDirectory #NTLM #RedTeam #Pentesting #PrivilegeEscalation #InfoSec

keyboard_arrow_left Previous keyboard_arrow_right Next

‼️ Google Awarded $250,000 Bounty for a Full-chain Chrome Exploit Found Using GPT-5.6 Details: cybersecuritynews.com/google-chrome-… Security researchers earned a $250,000 reward from Google after building a full-chain Chrome exploit with AI agents. The XBOW team says the work took six days and about $40,000 in computing costs, using OpenAI’s publicly available GPT-5.6 Sol model. Google has fixed the flaws. Its chain linked three bugs to move from JavaScript in a Chrome tab to code execution outside the browser’s renderer sandbox. The company estimated that the issue could affect about one billion users worldwide, while the demonstration targeted Windows. A sandbox limits what a compromised web page can do. A full chain joins bugs to cross Chrome’s defenses in sequence. #cybersecuritynews

Zero to Calc in Six Days: Inside XBOW’s Chrome Full Chain Author: B. DOLAN-GAVITT Source: xbow.com/labs/research/…

keyboard_arrow_left Previous keyboard_arrow_right Next

‼️ Microsoft Windows Shell 0-Day Vulnerability Actively Exploited in Attacks Details: cybersecuritynews.com/windows-shell-… Microsoft has patched an actively exploited Windows Shell zero-day vulnerability that allows attackers to bypass Mark of the Web protections. Tracked as CVE-2026-96951, the flaw was addressed in the October 2026 Patch Tuesday updates. It carries a CVSS score of 8.8 and is rated Important by Microsoft. The flaw involves command injection, where Windows Shell fails to properly handle special elements used in a command. Attackers can deliver a specially crafted file through email or instant messaging and use social engineering to persuade the target to open it. Exploitation requires user action. #cybersecuritynews #Microsoft

Exploit for CVE-2026-83834: github.com/wkpwn/CVE-2026… (my first windows exploit!!!!)

GLM-Legal uncensored:

Z.ai @Zai_org

2 weeks ago

Introducing GLM-Legal Built on GLM-5.3-Flash, GLM-Legal matches frontier models on mainstream legal benchmarks. Designed for real legal workflows: contract review, evidence analysis, compliance, and more. Tech Blog: z.ai/blog/glm-legal Apply for API Access:

340 573 8K 843K 3K

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 medium.com/@s12deff/steal…

If you're into Windows internals, malware analysis or low-level offensive security, bookmark this. @AliceCliment has a great FREE technical research archive covering: - Windows kernel internals - BYOVD + vulnerable driver research - Driver reversing + IOCTLs - Process-killer drivers - EDR bypass techniques - Import Address Table hooking + unhooking - Direct syscalls - Hell's Gate + Halo's Gate - FreshyCalls + SysWhispers2 - SSDT + syscall dispatch internals - NotPetya reverse engineering - Named pipes - Linux disk-encryption attacks Some particularly good rabbit holes: - Finding exploitable process-killer drivers using LOLDrivers - Reversing Avast + PowerTool kernel drivers - Following a syscall from user mode into the Windows kernel - Comparing syscall-ID retrieval techniques - Unhooking EDR modifications to the IAT - Reverse engineering NotPetya's named-pipe behavior A lot of it walks through the actual reversing process rather than just describing the technique at a high level. Start digging: alice.climent.red Huge credit to @AliceCliment for keeping the research publicly available. #WindowsInternals #ReverseEngineering #MalwareAnalysis

keyboard_arrow_left Previous keyboard_arrow_right Next

Reverse engineer anything with agents, from app behavior down to native binaries. github.com/morluto/rea

Publishing the "ApiHashing" repository based on the latest V2 update of the malware development course. The library replaces GetModuleHandleA/W, GetProcAddress, and GetModuleFileNameW and supports EAF bypass through ROP-style gadgets, forwarded functions, ordinal exports and API set DLLs versions 2 through 6. github.com/Maldev-Academy…

reverse-skill routes AI coding agents to the right reversing tool for a task, so they stop guessing between jadx, Frida, IDA, or Burp when faced with an app, binary, or CTF challenge. Explore it here: osp.fyi/reverse-skill