Unauthorized use of AI tools and AI-enabled applications introduces new attack surfaces, including data leakage, compliance failures, and AI-specific threats like prompt injection. Traditional security measures are insufficient to address risks from agent-to-agent interactions and unmonitored AI workflows.