NIST RFI: Security Considerations for AI Agents
Apollo Research's comment to NIST on agent security recommends scheming and misalignment evaluations, AI control and monitoring, and tighter oversight of high-stakes agentic deployments to prevent loss of control.