Identify and mitigate AI-specific threats, including prompt injection, jailbreaking, model poisoning, adversarial attacks, model inversion, unauthorized model access, excessive agent autonomy, sensitive data exposure, and AI supply chain risks. Establish and strengthen security controls for AI platforms, cloud environments, identity and access management, privileged access management, secrets management, data protection, monitoring, logging, model lifecycle security, and runtime enforcement capabilities.