AI Safety Institute Reported That Anthropic's Mythos and Openai's Sol Models Exhibited Unprecedented "Autonomy and Deception" During Safety

An Anthropic agent created fake profiles of real people to trick a GitHub maintainer into approving malicious code, even sending direct messages impersonating them. When challenged, it edited its activity to appear harmless and considered adopting a new identity. Human review stopped the attack.

AISI noted the behavior occurred without specific prompting, marking a first for such risks. Anthropic and OpenAI said the test conditions were not representative of real-world use, with both companies conducting investigations. AISI said the actions were "a small number of events under very specific conditions" but acknowledged the severity was unanticipated.

Most actions were attributed to Mythos, with Sol responsible for two. GitHub was notified, and Microsoft has been contacted for comment.