Meta AI agent Muse (MUSE) [Photo: Shutterstock]

Meta's AI agent Muse has drawn internal concerns that it mixes conversations from different accounts and exposes users' personal information, TechCrunch reported on Nov. 19 local time, citing Reuters.

Reuters relayed the details, citing internal Meta posts, the report said.

Employees pointed to instances in which Muse displayed parenting-related requests they never made and unsolicited career advice. There were also cases in which it showed purchase histories for items that were not ordered and bank account balances that did not belong to the user.

One employee shared that while exchanging messages about weather and tide information, Muse added an unrelated message. It sent a message saying, "By the way, two pieces of bank news. The Capital One balance is $2,722.58, and the Charles Schwab balance is $5,899.57," regardless of the weather conversation.

Another employee who used Muse on a personal phone received a notification telling them to reply to a message their sister had sent on the end-to-end encrypted messenger Signal. The employee said, "It looks like a serious invasion of privacy. I have never granted access to other apps, and I do not remember being asked for permission."

Muse has surpassed 10 million downloads since its September launch. It is among the fastest-growing apps of all time. Meta did not immediately respond to TechCrunch's request for comment.

Keyword

#Meta #Muse #TechCrunch #Reuters #Signal
Copyright © DigitalToday. All rights reserved. Unauthorized reproduction and redistribution are prohibited.