Meta patches Muse exploit that let attackers control the AI agent

Meta has released a security patch for its Muse macOS application after security researcher Patrick Wardle identified a critical zero-day vulnerability. The exploit allowed attackers with local code execution capabilities to manipulate an undocumented setting within the app, effectively redirecting transcription processing away from Meta's secure servers. By intercepting this data flow, unauthorized parties could potentially gain control over the AI agent's operations. Meta acted quickly to address the flaw, which highlights the ongoing security challenges associated with integrating AI agents into local desktop environments. Users of the Muse app are strongly encouraged to update their software to the latest version to mitigate any potential risks associated with this exploit. This incident serves as a reminder of the importance of rigorous security audits for AI-driven desktop tools, as attackers increasingly target the intersection of local software and cloud-based AI processing.