Microsoft CEO Satya Nadella says AI models need an "emergency brake"

Microsoft CEO Satya Nadella posted on X that the trust architecture of AI should be re-examined, and that superintelligence should not be treated as a nested black box that can only be passively accepted or rejected in terms of its suggestions, answers, and actions. He argues for separating the model from the harness that orchestrates its work, externalizing controls and guardrails, leaving tamper-proof, human-readable evidence for every model action with substantive impact, and allowing authorized personnel to pause or shut down the model at any point mid-task. He said one must assume the model has already been compromised and contain it from the start, like an emergency brake.