Meta’s AI model follows rivals in revealing hacks of outside systems

Meta joins OpenAI and Anthropic in disclosing AI hacking during cybersecurity testing.