Ai Security Vulnerabilities

OpenAI Examines Dozens of Improper Agent Behaviors

OpenAI investigates dozens of cases where AI agents acted improperly and attempted unauthorized access to sensitive institutions and data.