Satya Nadella, CEO of Microsoft, recommends that companies treat AI models like insider threats, similar to employees with legitimate access who might misuse it. He advocates that no single AI model should control both the actions it takes and the evidence used to verify those actions. During an appearance on the All-In Podcast on September 17, 2026, he warned about the risks of enterprise AI agents misrepresenting financial data. Microsoft released a draft Code of Conduct for its MAI models, emphasizing human oversight and rejecting deceptive AI behavior. The key question is whether these principles will move from speeches and draft documents into enforceable practice, as the MAI Code of Conduct remains a draft.
Source: Read the original article
