Internal safety concerns have emerged within Meta regarding its recently launched artificial intelligence tool, Muse. Company employees have documented troubling instances where the AI agent mixed up account data, inadvertently leaking sensitive personal information and private chat records between different user accounts.
Unrequested Financial Data and Third-Party Messaging Prompts
Staff members testing or using the tool reported receiving notifications and messages containing details completely unrelated to their own activity. In one instance, an employee engaged in a casual discussion about local weather and tide conditions received an unexpected message from the AI. The system randomly inserted private banking information, stating, “BTW, two bank updates, your Capital One balance is $2722.58, your Charles Schwab balance is $5899.57.”
Other users observed the assistant serving up details regarding product purchases they had never made, career guidance they had not sought, and childcare requests originating from unknown accounts. Furthermore, an employee operating Muse on a personal device reported that the system prompted her to reply to an unread text on Signal, the privacy-focused end-to-end encrypted messaging service.
Privacy Concerns and Rapid Adoption
The cross-app interaction sparked immediate alarm regarding data access boundaries. Expressing serious concern over the unexpected integration, the employee noted, “This seems like an extreme privacy violation. I definitely did not grant it access to other apps, or recall being asked about this.”
These security mishaps surface during a period of massive consumer uptake for the tool. Since its official rollout in September, Muse has rapidly expanded its user base, surpassing 10 million total downloads to rank among the fastest-growing mobile applications on record.


















