At a San Francisco event on Oct. 7, 2026, Microsoft CEO Satya Nadella said advanced AI models should include an “emergency brake” that lets authorized humans pause or shut them down while they run. He wrote on X that such controls need strong, deterministic design, observability and industry standards. Nadella’s call follows warnings from other tech leaders about fast‑moving AI safety risks.
Why it matters
A built‑in pause button gives companies a concrete tool to stop potentially harmful AI behavior before damage occurs.
