
Meta employees have reported potential privacy and security issues involving the company’s Muse AI agent, including conversations appearing to contain information from other accounts and unexpected references to private messages. According to a Reuters investigation published on November 18, 2026, internal company discussions revealed concerns that the assistant could expose sensitive user information or generate misleading personal details.
The reported incidents involved employees testing different versions of Muse, which Meta introduced in September as a personal AI assistant. The application has since recorded over 10 million downloads, making it one of the fastest-growing AI applications.
Employees Report Unexpected Personal Information
According to the internal posts reviewed by Reuters, employees encountered conversations containing details about childcare requests, career advice, purchases, and financial accounts that they did not recognize. Some employees questioned whether Muse was retrieving information belonging to other users or generating inaccurate responses.
In one reported incident, an employee asking Muse about weather and tide conditions unexpectedly received a message containing two financial account balances. The assistant identified a Capital One balance of $2,722.58 and a Charles Schwab balance of $5,899.57, despite the conversation having no apparent connection to banking.
Another employee reported receiving a reminder from Muse to respond to a message from her sister on Signal, an end-to-end encrypted messaging application. The employee said she had not authorized Muse to access other applications and could not recall being asked to provide such permission.
The employee described the unexpected reminder as a serious privacy concern. However, the Reuters investigation could not establish whether the reported financial details belonged to other users or were fabricated by the AI system.
Meta Responds to Internal Testing Concerns
A Meta spokesperson told TechCrunch that employee testing is an essential part of identifying problems before releasing new capabilities to consumers. The company said internal testers frequently use versions of Muse containing experimental features and functionality unavailable to the public.
Meta emphasized that the testing process allows employees to report unexpected behavior so developers can implement privacy safeguards and improve the assistant. The company also distinguished between the software versions used internally and those available to consumers.
According to Reuters, a member of Meta’s Superintelligence team suggested that some apparent account mix-ups were AI hallucinations caused by prompts encouraging Muse to initiate conversations between user sessions. Such behavior could produce fabricated messages that appeared to contain personal information.
The reported incidents come after Meta’s September launch of Muse, which introduced an assistant designed to help users manage daily activities through personalized interactions. The company has promoted the product’s ability to work with personal information and connected services while emphasizing privacy and security protections.
Featured image credits: Wikimedia Commons
For more stories like it, click the +Follow button at the top of this page to follow us.
