Meta Muse AI privacy issues have raised concerns after employees reported unexpected bank balances, private messages and other personal information.
A Reuters investigation published November 18 examined internal complaints about mixed conversations and unexpected access to unrelated applications.
Meta Muse AI privacy issues raise security concerns
One employee received unfamiliar bank balances during a conversation about weather and tides. Others reported unexpected childcare requests, career advice and purchase details.
Another employee said Muse referenced a private Signal message without her explicitly connecting the application.
Reuters could not establish whether the financial details belonged to other users or were incorrectly generated by the AI.
Tests reveal access through device permissions
Tests found Muse could retrieve Signal messages through Android notification permissions; on macOS, accessibility and screenshot permissions enabled access beyond explicitly connected applications.
Similar tests on an iPhone did not reproduce the behavior.
TechAmerica.ai previously covered Meta’s response to Muse privacy allegations.
Meta disputes widespread data exposure
Meta described the incidents as isolated. It said some apparent conversation mix-ups could be AI-generated errors rather than confirmed data leaks.
The company maintains that Muse requires user permission to interact with applications.
TechAmerica.ai also examined Apple’s AI agent security controls, including restrictions on application permissions and data access.
With more than 10 million downloads since September, Muse’s reported behavior highlights the importance of permissions and data isolation.










