Expert warns shadow AI and autonomous agents bypass security

T.J. Marlin, CEO of Guardrail Technologies, warned that AI agents with legitimate credentials look like authorized users, so identity-based defenses are no longer enough. He called for monitoring, containment and human intervention around agent actions rather than front-door controls.

• Marlin: front-door guardrails don't control what AI does next
• AI changes the economics of cybercrime, making attacks cheap and parallel
• Biggest shadow AI risk is invisible employee data leakage
• He urges monitoring and intervention before unsafe actions complete