Microsoft CEO calls for emergency brake mechanism for advanced AI models.

BlockBeats news, October 11 — Microsoft (MSFT.O) CEO Nadella stated that enterprises should treat powerful AI models as potential insider threats, assume they may have already been compromised, and establish "emergency brake" mechanisms to prevent autonomous agent models from spiraling out of control.

Nadella wrote on Saturday: "We must assume the models have already been compromised and impose constraints on them from the very beginning. Think of it as an emergency brake device. Authorized personnel should always be able to pause or shut down a model while it is executing tasks."

Nadella's security recommendations include: do not rely on a single model to make critical decisions; maintain tamper-proof records of agent behavior; and subject systems to independent audits. He also called for disclosure of major failures or security vulnerabilities, urging enterprises to share the causes of incidents and related details so that other companies can strengthen their security measures.