Unexpected call from Microsoft's CEO: Satya Nadella calls for the introduction of an “emergency brake” and independent controls for advanced artificial intelligence systems

Microsoft CEO Satya Nadella called on Saturday, in a message posted on the X platform, for the introduction of independent controls and an “emergency brake” for advanced artificial intelligence systems. The mechanism would allow authorized individuals to interrupt or shut down a model even while it is carrying out a task.

Nadella argued that nondeterministic models, whose actions cannot be anticipated in every situation, must be surrounded by deterministic systems, human oversight, and robust operational procedures. He proposed that AI systems, including proprietary systems and those with publicly accessible parameters, be treated from a security perspective as potential insider threats.

Among the recommended measures are continuous monitoring, easily understandable records of AI actions, ongoing testing, independent audits, isolation of dangerous components, and incident reporting. Nadella warned that safety must not depend solely on trust placed in a model, but on verifiable technical and institutional mechanisms.

The call comes amid mounting warnings about the accelerated development of artificial intelligence. Industry figures such as Sam Altman, Dario Amodei, Bill Gates, and Elon Musk have called for stricter safeguards, while the U.S. administration insists on maintaining the United States’ technological advantage over China.