Microsoft Warns Its Windows AI Feature Brings Data Theft and . . . Microsoft Warns Its Windows AI Feature Brings Data Theft and Malware Risks, and 'Occasionally May Hallucinate' (itsfoss com) 59 Posted by EditorDavid on Sunday November 23, 2025 @03:34AM from the game-of-Risks dept
Microsoft Warns its Agentic AI Features Can Be Hijacked to . . . Microsoft has admitted that its new Copilot Actions introduce "novel security risks" like Cross-Prompt Injection (XPIA), warning users that the feature can be exploited to install malware or steal
Windows 11 Agentic Features Trigger New Security Risks (MS . . . The novel security risks Microsoft calls out Microsoft uses unusually direct language in its documentation: agentic AI introduces “novel security risks ” The two most emphasized classes of risk are hallucinations (model errors) and cross‑prompt injection (XPIA), a practical evolution of prompt‑injection attacks in the context of agents that can act on their outputs Cross‑prompt
Microsoft Warns That Windows 11 AI Agents Could . . . - TheGamer Microsoft Warns That Windows 11 AI Agents Could Damage Your PC First reported by Windows Central (thanks Kotaku), a rather lengthy warning was recently published by Microsoft about its experimental agentic features that it's planning to add to Windows 11 relatively soon