AI

Meta's AI Assistant Muse Reports Mixing Up User Data and Breaching Privacy, Employees Say

Tendela Briefing ·
3 views · 0 Comments
Meta’s AI agent Muse is mixing up conversations and could be exposing private data, report says | TechCrunch

Image: TechCrunch · Source

Meta's AI assistant Muse mistakenly combines conversations and reveals incorrect private information such as banking details and personal messages, raising internal privacy concerns.

Meta's AI agent Muse, launched in September and rapidly adopted with over 10 million downloads, has come under scrutiny following employee reports of significant privacy issues. According to a Reuters investigation based on internal company posts, Muse has been observed mixing up conversations between different user accounts and revealing private data that does not belong to the user involved.

Employees reported instances where Muse surfaced unrelated information including childcare requests they never made, unsolicited career advice, product orders they had not placed, and even bank account balances belonging to other individuals. One example involved an unexpected message within a chat about weather and tides, which included specific bank balance updates for two different financial institutions. Another employee noted Muse sending a reminder about a message from her sister on Signal, a separate encrypted messaging app, despite no permissions granted to access that app.

Such incidents have understandably sparked serious concerns about user privacy and data security within the company. A Meta spokesperson responded by emphasizing the role of employee testing in product development, asserting that the versions of Muse tested internally often contain features and behaviors not yet available to the public. The testing process is intended to gather feedback for improving safety and privacy protections before wider release.

While Muse continues to grow in popularity, these reports highlight the challenges of deploying AI assistants that interact with sensitive personal information across multiple platforms. Meta faces the task of ensuring that Muse can deliver useful assistance without compromising user confidentiality or mixing data across accounts. The company has not disclosed specific technical causes behind these data mix-ups but acknowledges the need for enhanced controls.

As AI agents like Muse become embedded in everyday communication, incidents such as these underscore the importance of stringent privacy safeguards and transparency about data usage, especially during the developmental testing phases. Users and employees alike will be watching how Meta addresses these vulnerabilities to restore trust in their AI assistant’s handling of private data.

Sources and original reporting

Comments (0)

No comments yet. Start the discussion.

Write a comment

Comments are published after moderation. Your name and comment will be visible publicly. Account