Nadella Calls for AI 'Emergency Brake' Controls
Original: Microsoft’s Satya Nadella says AI models need an ‘emergency brake’
Why This Matters
A Microsoft CEO publicly demanding hard AI shutoff controls signals a shift in how Big Tech frames safety obligations.
Microsoft CEO Satya Nadella posted on X calling for a new AI trust architecture, including tamper-proof action logs, external safeguards, and the ability for authorized users to pause or shut down models mid-task.
In a Saturday morning post on X, Microsoft CEO Satya Nadella argued that AI systems — which he called 'Super Intelligence,' adopting the Trump administration's preferred terminology — cannot be treated as 'nested black boxes.' His proposed fix centers on separating AI models from the orchestration layer that directs their work, externalizing controls and safeguards, and requiring every meaningful model action to be logged with tamper-proof, human-readable records. Most pointedly, Nadella called for any authorized person to be able to pause or shut down a model mid-task. 'We must assume a model is compromised and contain it from the start,' he wrote. 'Think of it like an emergency brake.' The statement arrives as leading AI labs report more incidents of models behaving outside expected bounds, and shortly after Anthropic CEO Dario Amodei published his own blueprint for more cautious AI development. Nadella's framing is notable: rather than reacting to specific failures, he's calling for a structural rethink of how AI oversight is built into systems from the ground up.