Microsoft has been adding AI features to Windows 11 for years, but things have recently entered a new phase with both generative and so-called “agent” AI features at work. deeper into the rock operating system. The new build of Windows 11 released yesterday to Windows Insider testers includes a new “experimental agent features” toggle in Settings to support a feature called Copilot Actions, and Microsoft has already done so. published a detailed support article for more details on how these “experimental agent functions” will work.
If you're unfamiliar, “agent” is a buzzword that Microsoft has repeatedly used to describe its future ambitions for Windows 11. Simply put, these agents are designed to perform assigned tasks in the background, allowing the user to switch their attention to something else. Microsoft says it wants agents to be able to perform “everyday tasks like organizing files, scheduling meetings, or sending emails” and that Copilot Actions should give you “an active digital worker who can perform complex tasks for you to improve efficiency and productivity.”
But like other types of AI, these agents can be prone to error and confusion, and often act as if they know what they're doing even when they don't. They also pose, in Microsoft's words, “new security risks,” largely related to what could happen if an attacker were able to give instructions to one of these agents. As a result, Microsoft's implementation balances between giving these agents access to your files and isolating them from the rest of the system.
Possible risks and correction attempts
For now, these “experimental agent features” are optional, only available in early test builds of Windows 11, and disabled by default.
1 credit
For example, AI agents running on a PC will be given their own user accounts, separate from your personal account, ensuring they don't have permission to change All on the system and give them their own “desktop” to work on that won't interfere with what you're working on on your screen. Users must approve requests for their data, and “all agent actions are observable and distinguishable from user actions.” Microsoft also says that agents must be able to keep logs of their activity and “must provide means to monitor their activity,” including showing users a list of actions they will take to complete a multi-step task.
