Lates News

Microsoft (MSFT.O) CEO Nadella said enterprises should treat powerful AI models as potential insider threats, assume they may have already been compromised, and build "emergency brake" mechanisms to prevent autonomous agent models from going out of control. In a post on a social platform on Saturday, Nadella said: "We must assume the model has already been compromised and impose constraints on it from the outset. Think of this as an emergency brake. Authorized personnel should always be able to pause or shut down the model while it is executing tasks." Nadella's security recommendations include: do not rely on a single model to make critical decisions; keep tamper-proof records of agent behavior; and subject systems to independent audits. He also called for disclosure of major failures or security vulnerabilities, requiring enterprises to share the causes of incidents and related details so that other companies can strengthen their security protections.