AI
Microsoft's Nadella calls for an 'emergency brake' and a trust architecture for frontier AI
Microsoft chief executive Satya Nadella said advanced AI systems should carry an “emergency brake” that lets an authorised person pause or shut a model down mid-task, and that companies should treat frontier models as possible insider threats rather than trusted tools. In a 10 October post he argued deployments should never rest on a single vendor's assurances, proposing tamper-proof logs of agent actions, independent audits and prompt disclosure of major failures. The guidance lands after a run of disclosed incidents at OpenAI and Anthropic.