Microsoft CEO Nadella calls for ‘emergency brake’ on advanced AI models to ensure human control

Satya Nadella, CEO, Microsoft.Microsoft CEO Satya Nadella said companies deploying advanced artificial intelligence (AI) should be able to halt a model mid-task, proposing an “emergency brake” for such systems.His call comes amid growing scrutiny of AI risks, after Anthropic and OpenAI disclosed incidents in which models behaved unexpectedly.In a post on X, Nadella announced that we can’t treat Super Intelligence as a set of nested black boxes and simply accept or reject its recommendations, answers and actions.

He advocated building contained systems whose behaviour can be observed, limits that can be tested, and actions can always be contained.“We need to separate the supply of intelligence from the authority over it,” he said.Guardrails in placeNadella listed independent audits, tamper-proof records of model actions, the use of multiple models for critical decisions and mandatory disclosure of significant failures or security breaches as safeguards for frontier AI.His post also called out for a framework built around observability, model diversity, human oversight, continuous testing, independent controls, containment and incident disclosure.

It also championed the idea of prohibiting the models from executing tasks that violate their governing principles.