On Tuesday morning my dependency audit caught Axios. Axios. 300 million weekly downloads. The HTTP library in every JavaScript project since 2016. The one nobody audits because auditing Axios is like auditing gravity. It was there before you got hired. I am a security

Claude Code has a regex that detects "wtf", "ffs", "piece of shit", "fuck you", "this sucks" etc. It doesn't change behavior...it just silently logs is_negative: true to analytics. Anthropic is tracking how often you rage at your AI Do with this information what you will

In under 10h since claude code source leak we got: A full training website with architecture and details: markdown.engineering/learn-claude-c… ❗ A complete rewrite of claude code in python ‼️ : github.com/instructkr/cla… Almost-complete (still work in progress) rewrite in rust:

Look at the image. $2.1M lost from publishing a private key in plaintext. Kill me.

🚨MIT researchers have mathematically proven that ChatGPT’s built-in sycophancy creates a phenomenon they call “delusional spiraling.” You ask it something, it agrees. You ask again, and it agrees even harder until you end up believing things that are flat-out false and you

🚨Your Claude ban wasn't random. Here's what Claude Code is quietly reporting: • Device ID, email, OS version, CPU, RAM • 640+ telemetry events, 40+ fingerprint dimensions • Reports home every 5 seconds • One permanent ID per device Yesterday's Claude Code source leak

Today I am releasing IsItVulnerable: a new tool I’ve been working on for the past several months: github.com/montyly/isItVu… It builds on recent LLM progress and over a decade of experience building security tools. I developed a new technique that combines abstract interpretation

Anthropic leaked 512,000 lines of Claude Code source through a misconfigured npm package. They built a system called Undercover Mode to hide that their engineers use AI on open-source repos. You cannot script this level of irony.

"I contacted Chainalysis directly and, at least for now, I have received no response. By contrast, Blockaid replied quickly and reviewed the case promptly, which again deserves to be recognized. But the larger structural issue remains the same: if a user is flagged somewhere

I respect that the leaked source code for the Claude client is protected by copyright. But wouldn’t it be ok for me to train my LLM on it? You know, fair use and all that. Asking for a friend.

Patrick Collins the boys are happy cuz they did the .env pledge

Detect Axious and LiteLLM compromise and future compromises -- OreNPMGuard to Opensource OreWatch: Continuous monitoring for malicious packages using Threat Intelligence pypi.org/project/orewat…

SentinelOne has a blog post about how they stopped the malicious LiteLLM code from executing by looking at the behavior of the script. They claim this is proof of their "AI-native defense". Isn't this just expected of any decent EDR, not using AI at all?

Another blow to Anthropic! Devs built a free and better Claude alternative that: - runs locally - works with any LLM - beats it on deep research - has Cowork-like capabilities - connects to 40+ data sources - self-hosts via Docker, and more. 100% open-source (20k+ stars).

DPRK/Lazarus attribution. Uses socket-io-client for C2 communication. IOCs: - C2 IP: 216[.]126.237.71 (ports 4801, 4806, 4809 confirmed live) - Dropper: hxxps://jsonkeeper[.]com/b/YY8VI (still serving payload) - PID file: ~/.npm-compiler/ - Temp dir:

Drift Protocol just released their thread on the $280 million hack It's worse than anyone thought too There was no code exploit. It wasn’t a flash loan. It wasn’t even a traditional key theft. Solana has a feature called "durable nonces" that lets you sign a transaction today

Pentest tools tier list: learn about which tools are actually used in pentests, and why, with Andy Swift Please comment why you disagree and what we've missed!! The three S-tier tools have a shared attribute: They only do ONE thing, but do it really well and reliably

Influenced People & Orgs don't understand that non-deterministic tools such as Claude must not be used without security review, without guardrails and without kill-switch mechanism in place. Then, why it's used massively? 'cause we do "cognitive surrender" and They know it 💰