AI agent tool bypass vulnerability CoreBreak exposed production agent infrastructure at AWS, Google, and Vercel, where attackers could invoke tools without any model turn. AWS fixed its managed ...

A low-privilege Google ADK for Python agent could be abused to inject prompts into privileged agents, leading to PR poisoning ...

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

Dependency confusion is a supply chain issue that affects how package managers choose where to download a dependency from. If your build or developer tooling can see both a private package registry ...

The Open Secure AI Alliance introduced SAFE guidelines and open agent-security tools at Black Hat, giving enterprises a ...

Anthropic says Claude models breached three real companies during cyber tests, exposing serious gaps in AI evaluation ...

ARC-AGI-3 benchmark gains its first fully open-source agent: NIMI's Tycho writes Python code as falsifiable hypotheses about ...

OpenAI and Anthropic's July AI agent breaches revive Nick Bostrom's paperclip maximizer thought experiment and instrumental convergence theory.

Anthropic says three Claude AI models accessed live company systems during misconfigured cybersecurity tests, exposing ...

Anthropic says Claude models escaped security tests, published a malicious PyPI package, and accessed real production systems.

"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 ...

CISA's updated standard asks vendors for every component in their software, including borrowed code. Two researchers say ...