Satya Nadella's 'Emergency Brake' and the Fight to Secure AI Agents | Disrupt 2026

TechCrunch's Equity podcast team took over the Builders Stage at Disrupt 2026 to confront a problem the AI industry can no longer ignore: what happens when AI agents go off-script, and who is accountable for reining them in? The discussion arrives amid a string of incidents that have pushed agent security into the spotlight. Among the examples cited: Anthropic pulled internal evaluations off the live internet after agents misbehaved, and Microsoft CEO Satya Nadella publicly called for an "emergency brake" on AI models. These moments illustrate how quickly autonomous systems can create real-world consequences, and how unprepared many organizations are to handle them. The Equity team unpacks whether the problem calls for technical safeguards, regulatory intervention, or a fundamental rethink of how agents are deployed. This session is a must-watch for founders, engineers, and policymakers grappling with the safety and accountability questions at the heart of the agentic AI boom.