Amazon's Steve Schmidt on AI agents gone rogue (Live at HumanX) | Equity Podcast

Short summary

Amazon's Chief Security Officer Steve Schmidt discusses how AI agents are reshaping corporate threat landscapes. While external actors leverage AI to become dangerous at all skill levels, Schmidt emphasizes the greater risk: internal shadow AI and uncontained agents. He details Amazon's approach to agentic identity, traceability, containment architecture, guardrail security, and human-in-the-loop authorization, with specific security practices for startups deploying agents.

• •External threat actors are sharper with AI; the bigger risk is internal—shadow AI deployments and uncontained agents.
• •Amazon traces every agentic action to a human for accountability; containment and guardrails are critical attack surfaces.
• •Startups should implement human-in-the-loop authorization, know their assets early, and consider CISO resources based on AI deployment scope.

Generated with AI, which can make mistakes.