AUTO-UPDATED

Meta’s Muse AI Agent Has Been Downloaded Half a Million Times — With a Serious Bug

Meta has released a security patch for its new Muse AI agent on MacOS after cybersecurity researcher Patrick Wardle identified a critical zero-day vulnerability in the software.

Key Points

  • Cybersecurity researcher Patrick Wardle discovered a flaw allowing attackers to intercept Muse audio captures and access tokens.
  • Meta’s David Singleton confirmed the patch, though he disputed the researcher's claim that the exploit could be triggered remotely.
  • The Muse MacOS application reached over 500,000 downloads during its first week of availability.
  • Users are advised to update their Muse application immediately to ensure the latest security protections are active.
  • The incident occurred shortly before Meta Connect, where CEO Mark Zuckerberg is expected to discuss future AI hardware developments.

Why it Matters

This vulnerability highlights the significant security risks associated with granting AI agents deep access to personal data and system functions. As Meta expands its agentic AI ecosystem, the company faces ongoing pressure to prove it can protect user privacy and maintain digital security.
CNET Published by Katelyn Chedraoui
Read original