GoKawiilMeta's Muse AI gives false explanation of how it accessed user's Messages
A journalist testing Meta's Muse AI assistant on its new Mac app found it referencing contents of his Messages conversations, despite believing he hadn't granted access. When questioned, Muse gave a confused and inaccurate explanation, claiming it only saw notification previews synced through the paired Mac app. Meta executive David Singleton later clarified that Muse only accesses Messages data after a user explicitly opts in, and that the assistant's own explanation of its capabilities was simply wrong.
GoKawiil's interpretation of the reporting above, not reported fact.
The incident highlights a growing trust problem for AI assistants that are granted deep access to personal data: even when permissions are properly opt-in, users need the AI to accurately describe what it can see and why. An assistant that fabricates plausible-sounding but false explanations of its own data access undermines confidence in privacy controls, especially as these tools integrate more deeply into personal devices like Messages and Calendar.
- Meta's Muse assistant gave an inaccurate explanation of how it accessed a user's Messages data
- Meta says the underlying access was permission-based and opt-in, not unauthorized surveillance
- The episode reveals AI assistants can misrepresent their own internal workings, raising trust concerns
Source: theverge.com — Terrence O'Brien, 2026-09-19
Published there as: “Meta’s Muse is creepy, but maybe not for the reasons you think”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.