Meta patches Muse exploit that let attackers control the AI agent

Meta fixed a zero-day vulnerability in its Muse macOS AI agent that let attackers with local access redirect transcription data to their own servers, potentially controlling the AI and accessing user accounts. The exploit required local code execution but allowed actions like taking pictures and writing malicious files without user alerts. Meta patched the flaw promptly, emphasizing the low risk due to the need for local access.