EpochLex entry

Agent Security

AY-jent sih-KYOOR-uh-tee

The practice of protecting AI agents, their identities, sessions, tools, data flows, permissions, and execution paths from misuse, unauthorized access, prompt injection, data exfiltration, and other threats created or amplified by agentic behavior.