TechSatya Nadella’s ‘emergency brake’ and the race to secure AI agents
At Disrupt 2026, TechCrunch’s Equity podcast team discussed who is responsible when AI agents go off-script and how to keep them in check. Anthropic recently pulled internal evaluations from the live internet after agent misbehavior, while Microsoft CEO Satya Nadella called for an “emergency brake” on AI models, putting agent security in focus.









