Meta employees report privacy concerns as AI agent Muse leaks data

A glowing, star-shaped mechanical object floats against a dark background as Meta employees express concerns regarding data privacy. Illustration (AI-generated)

The Chronify

Share:

By Technology Desk

Meta employees have raised alarms that the company's AI agent, Muse, is leaking private user data and mixing up conversations between accounts. The tool, which has seen over 10 million downloads since its September launch, is reportedly exposing sensitive financial details.

Meta employees have raised significant concerns regarding the company’s artificial intelligence agent, Muse, alleging that the software is incorrectly merging conversations across different user accounts and exposing private personal data. The reports, which emerged from internal company discussions, suggest that the AI tool is surfacing sensitive information that does not belong to the users interacting with the system.

According to accounts shared by staff members, the AI has displayed a variety of personal details that users never requested or initiated. These include childcare requests, unsolicited career advice, and records of product orders that the employees had not placed. The nature of these leaks has prompted internal alarm regarding the security and integrity of the platform, which was launched by Meta in September.

In one documented instance, an employee reported that Muse unexpectedly interrupted a conversation about weather and tide information to provide specific financial data. The AI agent sent a message stating: "BTW, two bank updates, your Capital One balance is $2722.58, your Charles Schwab balance is $5899.57." The disclosure of such precise account balances has heightened fears among the workforce about how the AI processes and retrieves data from disparate sources.

Another incident involved an employee using the AI agent on her personal mobile device. The employee reported receiving a reminder from Muse to respond to a message from her sister on Signal, the encrypted messaging application. The staff member noted that she had never granted the AI application access to her other messaging services, nor did she recall being prompted to provide such permissions. In an internal post, the employee described the incident as an extreme privacy violation.

Since its release in September, Muse has experienced rapid growth and widespread adoption. It currently stands as one of the fastest-growing applications on record, having achieved more than 10 million downloads. The scale of the user base has intensified the scrutiny surrounding these reported technical failures, as the potential impact of such data exposure could affect a significant number of individuals.

Meta has not yet provided a formal response to inquiries regarding the reported technical issues. The company’s silence on the matter follows the internal circulation of these concerns, which highlight a potential breakdown in the data isolation protocols expected of such AI agents. As the investigation into these reports continues, the incidents remain a focal point for those monitoring the development and deployment of generative AI tools within the tech sector.

Elected News

Top Read News