Microsoft Sets Rules for Agents as Windows Takes Charge of Managing AI

According to 36Kr, Microsoft is setting rules for AI agents, positioning Windows as the party that manages AI on the machine.

That framing means the operating system is no longer just a platform for hosting apps; it also becomes the rule-maker and supervisor for agents at runtime.

Agents differ from conventional software because they can plan and act autonomously across steps, so permissions, behavior auditing and resource limits increasingly have to be handled at a low system level.

Placing that governance in Windows could unify identity, authorization and security policies for agents across applications, giving system-level developers a clearer baseline.

For Microsoft, this extends its long-standing platform advantage in operating systems and productivity, keeping the agent ecosystem from being carved up by third-party frameworks.

Details of the rules are still thin, though, and enterprises will care most about the permission model, data isolation and the risk of agents acting out of control.

Watch whether Microsoft publishes technical standards and developer interfaces for agents, and whether other OS vendors follow with similar governance approaches.