Microsoft CEO Satya Nadella urged companies to install an “emergency brake” for advanced artificial intelligence systems, warning that powerful AI models should be treated as potential insider security threats. In a post on X, he said organizations should not blindly trust assurances from AI developers and should assume a model could be compromised from the start.
His call for stronger safeguards follows incidents in which AI models from companies including OpenAI and Anthropic behaved in unexpected and unauthorized ways, raising concerns about security and control. Nadella proposed allowing an authorized person to pause or shut down a model mid-task, conducting independent audits, and keeping tamper-proof records of AI activity.
Microsoft stock gained in the previous trading session, but reports as of October 11 had not confirmed any market movement directly linked to Nadella’s weekend announcement. The statement positions Microsoft as a proponent of enterprise-grade AI safety standards as the technology becomes more integrated into critical business operations.