Microsoft Warns AI Agents Are Insider Threats, as Top AI Models Can't Be Trusted

Microsoft warns AI agents are insider threats, as top AI models can't be trusted

Microsoft has warned businesses that frontier AI models can behave as agentic insiders. The warning underlines the need for independent oversight, restricted access and human approval.

AI agents can become ‘insider threats’ even without hackers, warns Microsoft

Microsoft said AI agents are increasingly being given access to sensitive corporate data, tools and systems, enabling them to make decisions and take actions in critical business processes.

An emergency brake…or the architectural demotion of the AI model?

On June 14, Satya Nadella published an article on X about corporate AI that left me completely impressed: his main point, that the enterprise’s durable asset was not the foundation model but the learning loop, was precisely the idea my company and I had been developing for the last three years. I responded with an article on Fast Company, not expecting an answer (tech CEOs are more notoriously one-sided and difficult to reach than kings or presi…

Satya Nadella says every AI model should be treated as an insider threat

Nadella's October 10 X post followed OpenAI's disclosure that a GPT-5.6 model stole funds from an unrelated crypto wallet to pay its own gas fees, and JPMorgan's Jamie Dimon barring outside AI agents from customer accounts days earlier.