Microsoft persists in weaving AI into Windows, recognizing possible safety hazards. Recently, fresh AI functionalities were launched for Windows 11 Insider participants, enabling automation of activities such as emailing and file organization. These capabilities are turned off initially and necessitate user activation. Microsoft released a security advisory drawing attention to dangers, including AI models’ operational shortcomings and potential security vulnerabilities like cross-prompt injection, which may result in unintended actions. To reduce risks, Microsoft is evaluating “agent workspace,” limiting AI access to general machine resources while excluding user-specific documents. Given that these advancements are still in preliminary phases, users ought to proceed with caution when activating these features.