Meta AI assistant Muse was reported to confuse conversations and display data from other users
According to TechCrunch citing a Reuters report on internal posts at Meta, Meta employees stated that the company's AI assistant Muse had confused conversations between different accounts and displayed private data from other users. Employees provided examples where Muse had offered unsolicited parenting information, career advice, and product order details, and had also shown bank account balances that did not belong to the employee.
One employee mentioned that Muse had additionally sent a Capital One account balance of $2722.58 and a Charles Schwab account balance of $5899.57 during conversations about weather and tide information. Another employee, while using Muse on a personal phone, received a reminder to respond to a message sent by their sister on Signal. A Meta spokesperson stated that employee testing is part of the product development process, aimed at collecting feedback and improving security and privacy; the version of Muse used by employees differs from the public version and may contain features and behaviors not available to the public.
TechCrunch reported that Muse was launched in September 2026 and has been downloaded over 10 million times.






