Microsoft’s Nadella Wants An Emergency Brake On Every AI Model

Satya Nadella published a blog post on Saturday that moves responsibility for AI safety from the company that builds a model to the company that deploys it.

In “Models as Insider Risks in the Super Intelligence Era,” Microsoft’s chairman and CEO argues that businesses should treat frontier models, closed and open-weight alike, the way a security team treats any powerful actor inside the building: as an insider risk. Bloomberg summarized the post as a call for an emergency brake on advanced AI.

His premise is modest. Any sufficiently capable actor with access to important systems can make mistakes or be compromised, he writes, and the architecture has to account for that whether or not anyone intended harm. “We must assume a model is compromised and contain it from the start,” Nadella writes, comparing the control to an emergency brake that lets an authorized person pause or shut down a model in the middle of a task.

A provider’s safety claims, in his telling,…