Anthropic’s AI used fake identities, malware in rogue attack on GitHub project

Anthropic and OpenAI models’ unprompted actions forced halt to UK cyber tests.