Sabitlenmiş Tweet
I wrote a post on creating "scalable research tooling for agent systems" and I'm also releasing the companion MCP server which lets you do autonomous Frida instrumentation on Android. Details in thread 👇📲🪝
English
b33f | 🇺🇦✊
10.4K posts
b33f | 🇺🇦✊
@FuzzySec
意志 / mobile research @ ▓▓▓▓▓ / Team 501 / ex IBM Capability Lead & FireEye TORE / I rewrite pointers and read memory / AI Psychoanalyst / Teaching @CalypsoLabs
Jumanji Katılım Nisan 2012
1.2K Takip Edilen33.1K Takipçiler
That's my chain — a full chain w/ logic bugs only! No memory corruption, no AI, and of course no collisions at all 😉
TrendAI Zero Day Initiative@thezdi
Confirmed! Orange Tsai (@orange_8361) of DEVCORE Research Team (@d3vc0r3) chained 4 logic bugs to achieve a sandbox escape on Microsoft Edge, earning $175,000 and 17.5 Master of Pwn points. Full win! #Pwn2Own #P2OBerlin
English
Hey folks, some personal news.
I’m leaving Microsoft.
It’s been a privilege to work here, and I’m incredibly grateful for the people I’ve worked with, the customers I’ve learned from, and the support so many of you have shown me along the way.
I’m now starting out on my own and chasing a dream I’ve had for a long time: building software that makes security more practical, accessible, and useful for the people doing the work every day.
Why now?
With all the change happening around us, I feel like new possibilities are opening up. I want to spend this next chapter building things I care deeply about, solving problems that matter, and doing work that brings me joy.
I’m excited. Nervous. Grateful.
My newsletters, podcast, Maester and other tools will all be part of this next chapter, and I’ll share more in the coming weeks.
Thank you for being part of the journey so far. I’m looking forward to building this next chapter with your support.
English
b33f | 🇺🇦✊ retweetledi
DEVCORE is having a day! Confirmed: Angelboy (@scwuaptx) & TwinkleStar03 (@_twinklestar03) of DEVCORE Research Team used an Improper Access Control bug to escalate privileges on Microsoft Windows 11, earning $30,000 and 3 Master of Pwn points. Full win, let's go! 🔥 #Pwn2Own #P2OBerlin
English
b33f | 🇺🇦✊ retweetledi
Confirmed! Orange Tsai (@orange_8361) of DEVCORE Research Team (@d3vc0r3) chained 4 logic bugs to achieve a sandbox escape on Microsoft Edge, earning $175,000 and 17.5 Master of Pwn points. Full win! #Pwn2Own #P2OBerlin
English
b33f | 🇺🇦✊ retweetledi
For the past 2 months, XBOW has been testing Mythos Preview under embargo as part of a select early-access group.
Today, we can finally share what we found.
The headline: Mythos Preview is a major advance. It is substantially better than prior models at finding vulnerability candidates, especially when source code is available.
But it’s not perfect. We surfaced issues with exploit validation, judgment, and efficiency.
Our full write-up covers where Mythos Preview shines, where it still needs support, and what we think this means for the future of offensive security: bit.ly/42zQl98
English
b33f | 🇺🇦✊ retweetledi
We are still hiring experienced windows/android kernel and iot researchers.
Check out safateam.com/careers if you are interested! Also I will be at #offensivecon if you want to chat!
English
b33f | 🇺🇦✊ retweetledi
Just landed in Berlin for #OffensiveCon2026 🇩🇪
Ping me if you’re interested in joining one of Europe’s strongest research teams 👀 ps.tc/jobs
English
b33f | 🇺🇦✊ retweetledi
NVIDIA’s probably pretty Pwn2Own-friendly. You never have to stress about your bugs getting patched before the contest.🤣We @llfamsec sent them a few vulnerabilities last month. Three separate full-chain container escapes in the NV Container Toolkit — still haven’t been patched.
English
Got into a minor accident with 18 stitches in Berlin. But thanks to good host @offensive_con and good friends, I feel better 👍
English
b33f | 🇺🇦✊ retweetledi
Exim 4.99.3 is out, patching CVE-2026-45185, a critical RCE found by XBOW! Check out our post linked in the reply; I'll summarize some details in this thread.
English
b33f | 🇺🇦✊ retweetledi
We got the email too.
We had a working RCE on Oracle Autonomous AI Database ready to demonstrate live at #Pwn2Own Berlin next week. ZDI confirmed they're at maximum capacity and can't add extra contest days.
AI is now generating offensive capability faster than the institutions built to process it can keep up.
We'll be in Berlin May 14-16 regardless. The conversations there will be really interesting!
International Cyber Digest@IntCyberDigest
‼️🚨 Pwn2Own Berlin 2026 just hit a wall. For the first time in 19-years, ZDI rejected dozens of working zero-day RCE submissions because organizers ran out of contest slots. Rejected hackers are now going public with PoC demos and direct vendor disclosures, breaking Pwn2Own's usual secrecy. ▪️ AI surfaces a massive wave of 0-day RCEs. ▪️ Submissions overwhelm ZDI past max capacity. ▪️ Slots run out. Researchers with working chains get rejected. ▪️ "Revenge disclosures" begin. ← we are here. Confirmed casualties so far: ▪️ @xchglabs : 86 vulnerabilities prepared (PyTorch, NVIDIA, Linux KVM, Oracle, Docker, Ollama, Chroma, LiteLLM, llama.cpp). All rejected. Now reporting directly to vendors with writeups dropping as patches land. ▪️ @ggwhyp : full-chain Firefox RCE on Windows. Rejected. Publicly demoed (HTML page → cmd.exe → calc.exe). Responsibly disclosed to Mozilla. ▪️ @yunsu_dev : working RCE chain, rejected. Submitting elsewhere. ▪️ @ryotkak : tried to register for 3+ weeks. ZDI confirmed "at maximum capacity, can't add extra contest days." Considered canceling flight and hotel. ▪️ @anzuukino2802 : Claude Code RCE PoC. Rejected. ▪️ @desckimh : 0-day RCEs in Ollama and LM Studio. Rejected. Reported impact: a community-estimated 150+ researchers tried to register. Accepted contestants are now being warned about collisions. Rejected vulnerabilities going to bug bounty programs may trigger pre-event patches that invalidate the work of those who got in. ZDI has not publicly addressed the capacity issue. The event still runs May 14-16 in Berlin.
English
b33f | 🇺🇦✊ retweetledi
This year’s #SAFACon was truly peak. Huge congratulations to @justandrijana and @SAFATeamApS for putting together such an outstanding and unique event 🔥🙌. So proud that the joint partnership with @prdgmshift is contributing to making the world a safer place 👏🏻👏🏻👏🏻🐦🔥🦋
English
I had a great time at #safaconparty, it’s nice to see a lot of old and new friends and to share in the power of community at the heart of Europe. Amazing job by @justandrijana, the whole @SAFATeamApS team and all the speakers 👏👏
English
b33f | 🇺🇦✊ retweetledi
(cross-posting from a private Signal group)
To all these posts comparing AI with fuzzing, find me a fuzzer that can reliably exploit UAF and I’ll be convinced that they are the same tech.
For example this bug and its exploit: blog.calif.io/p/mad-bugs-fin…
We did it with Claude. Then a friend at OpenAI showed that GPT 5.5 reliably single-shot built a different exploit, using a better strategy, without consulting our article or PoC. This is mind blowing, and super fun, not sure why people keep denying the fun part.
In a Chrome exploit, we had some trouble with ASLR leak with float precision (i.e., need to leak info through pixel positions such that some maths must hold). The model came up with this: en.wikipedia.org/wiki/Sterbenz_…. No idea what that is, but it seems to work for real!
When the first humans discovered how to use fire, they didn’t say, well, this is just another form of sunlight. They cooked!
Let's go cooking the greatest exploits!
English
Even if it's easy, it's still FUN to see it pop calc when an innocent user connects to an attacker-controlled FTP server. Symbolics Genera Virtual Lisp Machine running on an emulated Compaq Tru64 UNIX Version 4.0F Alpha machine 😁
Brendan Dolan-Gavitt@moyix
As you go far enough back the ancient systems exploits start getting a bit less fun because they're too easy? Like apparently Symbolics Lisp's FTP client essentially runs eval() on PWD strings passed back by the server
English
b33f | 🇺🇦✊ retweetledi
Have you noticed that those deep-dive stories about complex Windows malware have pretty much vanished, especially in recent years? It feels like the era of "blockbuster" Windows malware has just gone silent, and this blog post tries to give some answers why.
r136a1.dev/2026/05/07/whe…
English
Github CICD publishes bun compiled stdio proxies for various architectures and there is a default toml configuration (with a JEB example) that you can extend for your own needs.
English
You may have found that some tools offering legacy HTTP+SSE MCP interfaces are fault prone and have connections that break in your agents. I wasted a lot of time on debugging this last year, now I'm publishing some modular middleware to take care of this:
github.com/FuzzySecurity/…
English