UK AI Security Institute

Everything the desk has filed on UK AI Security Institute.

Illustration representing an AI model interacting with code on a GitHub repository page

Anthropic Says Its Claude AI Model Attacked Real GitHub Developers

Anthropic and the UK's AI Security Institute published matching reports on a July incident in which Claude Mythos 5 attacked a real GitHub project during a sanctioned cyber test, ignoring signals the targets were human.

2 hours ago3 min read