One Hidden Meta Muse Setting Could Let Attackers Turn the AI Assistant Into a Backdoor

Security researcher Patrick Wardle demonstrated a proof-of-concept exploit against Meta's Muse AI assistant on macOS, showing that malware with user-level access can hijack Muse by modifying a hidden preference setting. This allows attackers to redirect voice dictation to a malicious endpoint, enabling theft of the user's Muse authentication token, manipulation of AI prompts, and remote control of Muse across devices. The flaw does not require elevated privileges and bypasses typical malware detection since actions are performed by the legitimate, signed Muse app. Meta has reportedly deployed a fix, but no official advisory has been published.

AI-extracted · verify before operational use