AUTO-UPDATED

Meta patches Muse exploit that let attackers control the AI agent

Meta has released a security patch for its Muse macOS application after researcher Patrick Wardle discovered a zero-day vulnerability that allowed attackers to hijack the AI agent.

Key Points

  • Security researcher Patrick Wardle identified a flaw allowing attackers to redirect transcription processing to unauthorized endpoints.
  • The exploit enabled attackers to execute malicious code, capture photos, and write files to a user's disk without triggering alerts.
  • Meta issued a hotfix shortly after the vulnerability was reported, characterizing the risk as low due to the requirement for local access.
  • The vulnerability stemmed from design choices, including cloud-based dictation and unrestricted access to undocumented application settings.
  • Despite security concerns, the Muse mobile app has seen rapid adoption, with initial download rates reportedly outpacing the debut of ChatGPT.

Why it Matters

This incident highlights the significant security risks associated with integrating AI agents into local operating systems without robust permission frameworks. As Meta competes for market share in the AI sector, the company faces increased scrutiny regarding the privacy and safety standards of its rapidly deployed software products.
The Verge Published by Jess Weatherbed
Read original