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

Security researcher Patrick Wardle demonstrated a vulnerability in Meta's Muse AI assistant for Mac that allows malware with user-level access to redirect voice dictation to an attacker, effectively turning Muse into a backdoor with broad access to user data and devices.
- Un-Mused: How a Single Debug Setting Bypassed macOS Security in Meta’s AI Client· InfoQ (AI, ML & Data)
