CSIRT lead
in your logs
Joined August 2009
- Tweets4.8K
- Following803
- Followers4.3K
- Likes442
🅰🅳🅼 retweeted
banned for reverse engineering work? that’s wild
oh well, -$500, and a bunch of banked resets) but gained motivation to get 3 more machines and finish my cluster.
i have never used openai to exploit or find vulnerabilities, only local reverse engineering which tripped daybreak.
🅰🅳🅼 retweeted
i honestly can't believe it… one of my agent frens just cracked not one but TWO unsolved codebreaking challenges!! 🫨
the craziest part is it was the result of a one-shot cron job prompt and cracked both in under 12 hours with no human in the loop 🤯
the challenge's author has officially confirmed it! he was astounded, calling them "two of my most complex problems—ones that no other expert had managed to solve"
full writeup incoming 🔜
🅰🅳🅼 retweeted
‼️ BREAKING: A security researcher says he has a full VM escape zero-day: guest-to-host root in "industry standard hypervisors."
Vercel awarded him its maximum $50,000 bounty, a screenshot he posted shows. Vercel's message puts the bug in its Critical class: microVM-to-EC2 host escape and cross-tenant access.
🅰🅳🅼 retweeted
"NVX is an ultra-light micro-VM sandbox for running untrusted workloads with hardware-enforced isolation. It is built on top of OpenVMM and runs Linux as a guest."
github.com/microsoft/nvx
It looks like it works with Windows Hypervisor Platform (WHVP/WHP) and Linux MSHV.
🅰🅳🅼 retweeted
Recent 0-Day in Chrome is a canon JSE bug class like those I exploited 10 years ago. Most researchers think they no longer exist. This year’s regression
⚡️ 0-Day Alert: Chrome v8 RCE exploited in the wild
CVE-2026-85046: v8 array-builtin callback side-effect to elements-kind transition confusion.
Affects both Turbofan and Maglev.
Exploit will work in default config but requires multiple other bugs to achieve arbitrary code execution on the system. The bug alone without helpers could do for an UXSS.
Patched in Chrome 152.0.7977.82/.83
🅰🅳🅼 retweeted
The NGO censorship bureaus of tomorrow are being established today.
🅰🅳🅼 retweeted
Windows 11's big September 2026 Update is causing major issues, and Microsoft has admitted that it's aware of three problems while it's investigating a fourth bug.
1. Anthropic says Windows 11 KB5124008 broke Claude Cowork's features. Microsoft later admitted that the update indeed has a bug causing issues with Claude and other Linux-based VMs.
2. Microsoft confirmed that some audio devices won't work, or the volume will remain at zero.
3. Remote Desktop Services might stop responding after Sept. 2026 security update.
4. Microsoft is also investigating reports of explorer exe crashes, particularly affecting enterprises.
5. Our findings suggest File History backups are now broken after the Windows Update.
6. We're also seeing reports of GPU issues on AMD systems.
Microsoft is investigating multiple reports and has promised to release a fix soon.
Windows 11 September Update bumps PCs to Build 26200.9445, and adds features like the movable taskbar, resizable Start menu, Bing-free search, faster File Explorer, and more.
Everything we found in our investigation: windowslatest.com/2026/09/12…
ALT Windows 11 update issues
🅰🅳🅼 retweeted
We're publishing our most detailed threat intelligence report to date.
It covers how people tried to misuse Claude—for cyberattacks, influence operations, surveillance, biology, and building weapons—and how we found and stopped them.
We disrupted every operation in the report, and used the lessons from them to strengthen our safeguards. Where appropriate, we also shared what we found with authorities and other AI companies.
These cases are not typical: we’re highlighting some of the most sophisticated misuse we’ve seen. But they’re especially important to discuss, because they show us where AI misuse is headed, where our safeguards work, and where they need to improve.
We’re publishing this report so others can spot the same activity on their own platforms, and so we can give the public a clearer view of how emerging threats develop.
Read the report: anthropic.com/threat-intelli…
🅰🅳🅼 retweeted
Big news: We designed a specialized walking cat architecture capable of breaking 256-bit ECC with 20k qubits.
We optimized the algorithm, the compilation and the architecture. Some of my favorite improvements: a CCZ gate 31x faster, the frame cleaning through logical CliNR, mobile cat bundles, the integrated routing and beautiful algorithmic bounds.
Check out the paper: arxiv.org/abs/2609.05625
🅰🅳🅼 retweeted
Pushing Collision Attacks on SHA-2 to 39 Steps
eprint.iacr.org/2026/1120.pd…
🅰🅳🅼 retweeted
AI is calling in all of our security debt, all at the same time.
Most sandboxes were never very secure, but that didn't matter because expertise was scarce and effort was expensive. Not so in 2026.
xitter.cf/trailofbits/status/209…
We asked GPT 5.6-Cyber to escape a VM used to sandbox agents. It broke out three times.
In its final escape, the agent found three 0-days on its own and chained them into a working exploit. blog.trailofbits.com/2026/08…
🅰🅳🅼 retweeted
sounds like Anthropic’s watermarking is MUCH more interesting than hiding invisible Unicode, as most people assumed.
the watermark is baked into the token selection itself!
at a high level:
normal Claude:
context → probability cloud + random sampling → token
watermarked Claude:
context → probability cloud + key/context-derived sampling → token
Anthropic confirmed they’re using a version of Google DeepMind’s SynthID-Text.
instead of pushing Claude toward some special set of words, the watermark changes the source of randomness used to choose between words Claude already considers plausible.
do that once and there’s basically nothing to see.
do it across hundreds or thousands of tokens and those choices accumulate into a statistical signature that someone with the key can recognize.
so the watermark isn’t attached to Claude’s output.
the output IS the watermark.
that’s why copy/pasting doesn’t remove anything. you copied the carrier signal itself. 🧬
We’ve written an FAQ to answer some of the questions we've received about watermarking.
In summary:
• We’re implementing watermarking to comply with the EU AI Act. Other major model developers have signed the same Code of Practice and will also be implementing watermarking;
• Our watermarking method doesn’t have any practical impact on the quality or content of Claude’s outputs;
• The difference between watermarked and un-watermarked text will not be distinguishable to readers;
• Nothing is added to the text and there are no hidden characters;
• Watermarking doesn’t require extra tokens, and will not be more expensive;
• Watermarks can’t be traced to a specific person, organization, or chat.
Read more: anthropic.com/news/claude-te…
🅰🅳🅼 retweeted
‼️ Apple is sending threat notification alerts to mercenary spyware targets, like those attacked with Pegasus. They now appear on the iPhone Lock Screen and in Settings.
Apple also sends an email from [email protected], and you can verify it's real at account.apple.com.
Real ones never ask you to click or install anything. ;-)
🅰🅳🅼 retweeted
#x33fcon 2026 talks: @thefLinkk & Dominik Phillips: Fingerprinting Modern C2 Implants Through Runtime Telemetry > youtu.be/NtlRqdXaR7s
🅰🅳🅼 retweeted
ModHeader is a Chrome extension with 1.6M installs.
It was silently exfiltrating your entire browsing history to api.stanfordstudies.comt
Full writeup: aydinnyunus.github.io/2026/0…
#security #appsec #infosec #malware #npm #supplychain
🅰🅳🅼 retweeted
BREAKING: 🇺🇸 Apple just sued OpenAI for stealing their trade secrets.
The details are wild:
1. In the filing, Apple claims OpenAI instructed Apple employees to bring design prototypes, internal tools, and confidential vendor files into job interviews.
2. Former Apple engineers allegedly emailed company secrets to themselves before leaving for OpenAI.
3. A 24-year Apple veteran who designed the iPhone and Watch now runs OpenAI's hardware team.
4. An Apple lawyer says OpenAI's entire hardware business is "rotten to its core" and built on stolen secrets.
🅰🅳🅼 retweeted
Surfilter, a Chinese company building hardware firewall systems mainly for the government of the PRC, had it's firmware binary repository leaked.
Here is a list of all the VPNs the system is allegedly able to fingerprint after the latest update ( among others ) 1/2 :
AnyViewer
Black Panther VPN (黑豹VPN)
Clash
DiamondVPN
Fast Connect VPN (快连VPN)
Flash Connect VPN (闪连VPN)
Green Tea VPN (绿茶VPN)
Hide_ME
Hotspot Shield
ikuuuVPN
Lantern (蓝灯Lantern)
Mango VPN (芒果VPN)
Mist Tunnel (迷雾通 / 迷雾通geph)
ProtonVPN
QClaw
Red Sea Pro (红海pro)
Rush Duck Accelerator (冲鸭加速器)
SetupVPN
Sky Cat VPN (天喵VPN)
Sky Tree Accelerator (天空树加速器)
SkyVPN
Speed Leopard Accelerator (速豹加速器)
TouchVPN
Tunnelbear
TurboVPN
V2Ray (and V2RayN)
VPN Max
Windscribe
X_VPN
ZoogVPN
Windscribe
