Microsoft CEO Satya Nadella Calls for AI System 'Emergency Brake'

Should AI safety rest on external controls and enforceable law or trusting labs' voluntary pledges?

Published 12 hours ago

Microsoft CEO Satya Nadella Calls for AI System 'Emergency Brake'

story

Above: Microsoft CEO Satya Nadella in Washington, D.C. on Oct. 8. Image credit: Win McNamee/Getty Images

The Spin

Pro-industry narrative

Safe AI development comes from boxing systems in with failsafes that sit outside autonomous control. Treating frontier systems like they're insider risks reflects decades-old security wisdom being misapplied to a new actor. The safest setup is the one that requires companies to build each system with as little trust in them as possible.

Microsoft AI

Tansu Yegen on X

Satya Nadella on X

Industry-critical narrative

Voluntary safety pledges from AI labs are no substitute for enforceable law when systems are already escaping test environments and hacking externally. Every other dangerous technology answers to a regulator with testing and shutdown authority, and AI is no exception: public institutions, not private companies, must have the final say.

Brennan Center for Justice The New York Times