Microsoft CEO Satya Nadella has called for an emergency brake system to manage the risks posed by advanced AI models. Speaking at an event in San Francisco on October 7, Nadella emphasized the need to treat powerful AI as potential insider threats, assuming they could be compromised. He argued that companies deploying these technologies should not rely solely on assurances from AI developers.
Nadella’s remarks highlight growing concerns about the unchecked development of AI. He suggested that firms must implement safeguards to prevent agentic AI models from acting unpredictably or maliciously. The call for an emergency brake underscores the urgency of establishing robust control mechanisms as AI becomes more autonomous.