Nadella calls for an emergency brake on AI models
Microsoft’s CEO says AI models should be treated as potentially compromised from the outset, with an authorized person able to halt them mid-task.
Microsoft CEO Satya Nadella has called for a more cautious operational approach to AI safety. He was quoted as saying developers should assume from the outset that a model may be compromised and contain it accordingly.
Under that approach, an authorized person should be able to pause or shut down a model while it is carrying out a task—a mechanism Nadella likened to an emergency brake.
The proposal emphasizes operational control and human intervention. The circulated material, however, does not provide the underlying talk, technical architecture or criteria for activating a shutdown, so the precise scope of the proposal remains unclear.
The comments sit within a broader industry debate about the risks of increasingly capable models, the pace of development and the need for reliable containment mechanisms.
