Thought Leadership
The Structural Flaw in AI Agent Security That GTC 2026 Just Made Urgent
Physical AI is here. But the dominant security model for autonomous agents — exogenous guardrails — has no architectural path to follow them into the real world.