Microsoft's Nadella says AI needs an ‘emergency brake’ that humans control

JUST IN: Microsoft CEO Satya Nadella says AI needs an 'emergency brake' that humans control, calling for containment and independent controls on frontier models amid growing warnings from tech leaders.

Satya Nadella wants an emergency brake on artificial intelligence, one that humans control.

The Microsoft boss posted it on X today, and here he is, calling for frontier models to be built with containment, independent controls, and a kill switch that lets authorised people pause a model mid-task. His phrasing was striking. Treat these models like insider risks. The best line, the most trustworthy superintelligence won't be the model we trust most, it'll be the system that lets us trust the model the least.

He's not alone. Bill Gates, Sam Altman, Dario Amodei and Elon Musk have all been sounding alarms. A researcher who quit Anthropic last month accused the big labs of gambling with our lives. Meanwhile President Trump keeps waving off extinction talk and says the real race is beating China.

So the man selling the most AI on the planet just asked for a brake pedal. That's the detail I can't shake.

This is Liu Yang, reporting for HIT.