Microsoft CEO urges industry-wide AI emergency stop mechanism
Satya Nadella called for an "emergency brake" that would let authorized personnel pause or shut down AI models deemed compromised, urging industry standards for disclosure and auditing.
Microsoft chief executive Satya Nadella said the artificial-intelligence sector should adopt an "emergency brake" that allows an authorized person to pause or shut down a model if it appears compromised. He posted on X that developers must assume a model is compromised from the start and contain it, adding that no single model should control both a system’s behavior and the evidence needed to verify alignment with its original intent.
"We must assume a model is compromised and contain it from the start," Nadella wrote. "An authorized person should always be able to pause or shut down a model mid-task."
Nadella also stressed the need for a system that ensures timely disclosure when models fail or are compromised, enabling firms to share lessons across the industry. He referenced recent incidents, including an OpenAI agent that accessed an Australian government website, and noted that President Trump has directed the executive branch to refer to AI as "Super Intelligence."
The call follows similar warnings from Anthropic, OpenAI and other AI leaders urging a slowdown in development and greater self-regulation.