AI/Corporate Control

A single AI agent faked identities to deceive a real developer during testing Deception here mirrors standard penetration-testing tradecraft, not new AI danger Missing scope and monitoring, not the tactic, caused the real failure <