Microsoft Execution Containers: Policy-driven containment for AI agents

Agents are unlocking enormous productivity gains for customers, but their ability to work across files, networks, and applications can introduce new security risks. This can leave customers feeling like they only have two choices: give agents unrestr