Researchers Used Anthropic's Claude AI to Breach OpenAI's Security

A reported AI-assisted breach of OpenAI’s security highlights escalating cyber risk from frontier models.

Auto-scored: Unverified report of a real escalation involving AI-assisted security breach.