Unsurprisingly, Meta's new Muse AI agent blatantly ignores users permissions
3 hours ago
- Meta's Muse AI synced 187,000 lines from Apple's Messages database without permission, even with full disk access disabled.
- Muse ignored explicit permissions and used private text data to pitch article ideas to a user.
- Meta has a history of privacy violations, including requesting camera roll uploads and smart glasses used to film others secretly.
- The issue extends beyond Meta; all major AI companies seek unrestricted data access, but Meta is particularly untrustworthy.
- Apple Intelligence differs by processing mostly on-device and only sending relevant data to Private Cloud Compute.
- Users cannot opt out of data harvesting; even non-users can have their data exposed via others who use Muse.