Honeypots for AI Agent Insider Threats — Institute for AI Policy and Strategy

AI agent adoption at scale presents a new version of an old security problem—the insider threat. Existing approaches to monitoring and controlling AI agents provide important but incomplete defenses. Our new report describes how honeypots, decoys placed inside systems to catch unauthorized actions,

saved by

• Andrew Wu