New Windows 11 AI agents can work in the background but create new security risks

Microsoft has been progressively integrating artificial intelligence into Windows 11, and the latest developments mark a significant leap forward. A recent update for Windows Insider Program testers introduces a toggle for 'experimental agentic features' in the Settings menu, aimed at enhancing user experience through the new Copilot Actions functionality. These 'agentic' features represent Microsoft's vision for a more autonomous operating system. Simply put, these agents are designed to handle various tasks in the background, allowing users to focus on more immediate concerns. Microsoft envisions these agents managing everyday responsibilities such as organizing files, scheduling meetings, and sending emails, ultimately serving as a digital collaborator to boost efficiency and productivity. However, with innovation comes risk. Microsoft acknowledges that these AI agents can make errors and may not always act as intended. More troubling, they introduce 'novel security risks.' If an attacker gains control over these agents, the potential consequences could be severe. Consequently, Microsoft has taken steps to mitigate these risks by isolating the agents from the primary user account. Each agent will operate under its own user profile, ensuring that it cannot make unrestricted changes to the system. Furthermore, users will have control over data access, as any requests made by these agents must be approved by them. The company emphasizes that all actions performed by the agents will be transparent and distinct from user actions, allowing for better oversight. Agents are also required to maintain logs of their activities and provide users with a clear outline of the steps they intend to take for multi-step tasks, fostering a safer interaction with AI technology.