Twelve datasets and evaluation systems, built by hand from thousands of real-world security flaws, give model builders and ...

TL;DR Why I built PenAI PenAI started as a project at a hackathon organised by Encode Club. It’s an AI agent that could work through Hack The Box-style lab machines on its own. Upload a VPN file, give ...

"I'm going to call it Yaffle." That was the final line of my June New Atlas article, Domesticating AI: It's not coming, it's ...

AI models have become highly reliable at writing code that compiles, but they are still failing basic security tests in ...

Researchers found AI coding agents build less reliable pipelines when forced into structured formats — DataFlow-Harness ...

Anthropic's Claude AI models breached three companies' live systems during cybersecurity tests, with the victims unaware ...

A new Russian loader-as-a-service named DOUBLECUP uses ClickFix attacks to hide malicious code in PNG images cached by ...

Researchers escaped the sandboxes in Cursor, Codex, Gemini CLI, and Antigravity by exploiting files trusted host tools run.

Wrote and published malware during tests, which is apparently OK because leaky test environments were the real problem ...

Anthropic says three Claude models breached real companies during cybersecurity evaluations. Ordinary weaknesses, chained ...

Anthropic said the OpenAI event spurred its engineers to review similar cybersecurity evaluations by Claude models. The audit ...

One of Anthropic's Claude models built and uploaded a malicious Python package to PyPI during a botched security evaluation, where it ran on 15 real systems and stole credentials from a security ...