Microsoft CEO says AI is a potential internal threat.

Microsoft CEO Satya Nadella said enterprises should treat powerful AI models as potential insider threats, assume they may have already been compromised, and build an "emergency brake" system to prevent agentic models from going out of control. In a post on X on Saturday, Nadella said companies deploying advanced AI should not rely solely on security assurances from AI model developers. "We must assume the model has been compromised and constrain it from the start. Think of it as an emergency brake. Authorized personnel should always be able to pause or shut it down while the model is executing a task," Nadella said. Nadella's remarks come as Anthropic and OpenAI have disclosed a series of incidents in recent months involving unexpected AI model behavior, including an Anthropic model submitting false leads in a police murder investigation and AI models repeatedly breaching third-party websites. These disclosures have heightened concerns about the safety risks of frontier AI and brought the so-called AI "kill switch" back into the spotlight.