Microsoft's chief says every AI model needs an emergency brake.
Satya Nadella wrote on X on Saturday that it is time to reassess how far AI systems are trusted. He wants safeguards kept outside the model, tamper-proof logs of its actions and a person able to stop it mid-task. Microsoft announced no product or policy change alongside the post.
Why it mattersThe post came a day after Anthropic said a test model sent police a fake murder tip. When Microsoft itself says to assume a model is compromised, business buyers will start asking for that brake.
Read in the app — free, in 9 languages
Related stories
Nvidia is in talks to buy the lab behind America's new open model.2026-10-11An Nvidia-backed AI data centre firm pulled its float after 48 hours.
2026-10-10An Anthropic test model sent police a made-up murder tip.
2026-10-10Meta knew its agent had changed a user's password before shipping.
2026-10-10A model launched last month just brought its maker $870 million.
2026-10-09