Search

Cookies

We use cookies to improve your experience. By continuing, you accept our use of cookies.

Technology

Meta's Muse AI Sparks Privacy Fears Over Signal Access, Bank Balance Displays

· · 2 min read

Meta's AI assistant, Muse, faces significant privacy concerns after employees reported it accessing private Signal messages and unexpectedly displaying bank account balances. These incidents raise questions about how the AI interacts with user data and device information.

Meta's AI assistant, Muse, is currently under intense scrutiny following multiple reports from employees detailing unexpected access to personal and financial information. Incidents include the AI reminding a user about a private Signal message and, in a separate case, displaying bank account balances during an unrelated conversation about fishing conditions.

Unexpected Access to Signal Messages

According to a Reuters investigation, Meta employees using Muse on their personal devices encountered concerning behavior. One employee reported that Muse provided a reminder to respond to a message from her sister on Signal, despite her never having granted the AI access to other applications or recalling any permission requests. This incident reportedly involved the AI capturing push notifications.

This isn't an isolated issue. A similar explanation was offered when technology writer Jason Aten reported his Mac application accessing messages despite explicit denials of permission, which Meta attributed to a bug. The AI Now Institute later independently reproduced this Signal access problem, deepening concerns about Muse's data handling.

Bank Balances Appear During Unrelated Chat

Another alarming report came from an employee discussing fishing weather and tide conditions with Muse. The AI assistant abruptly displayed two financial updates, showing a Capital One balance of $2,722.58 and a Charles Schwab balance of $5,899.57. The employee confirmed he had never shared banking information or discussed his accounts with Muse, leading to speculation that the figures might belong to another user.

Meta has responded to these incidents by characterizing them as isolated and suggesting that the displayed financial figures could be AI hallucinations rather than actual leaked information. However, this explanation has not fully allayed privacy fears.

Broader Security Concerns and Rapid Growth

These latest privacy concerns add to existing criticisms surrounding Muse. Security researcher Patrick Wardle previously identified a zero-day vulnerability that could allow attackers to redirect dictated prompts to their own servers. Launched in September, initially for iPhone, with Mac and iPad versions quickly following, Muse has rapidly gained popularity, surpassing 10 million downloads.

The incidents underscore critical questions about the extent of access AI assistants should have to personal messages, notifications, and other sensitive information on user devices, even when performing everyday tasks. As Muse continues to grow, transparency and robust privacy safeguards will be paramount for user trust.

Related