AI Security, Safety & Agent Control

AI-specific security threats and safety incidents: prompt injection, model/sandbox escapes, agent identity, permissions and containment, AI-enabled cyberattacks, and privacy risks created by AI systems.