Meta’s Muse AI Agent Caught Leaking Users’ Intimate Personal Information
"We would have guessed that gathering these would take real effort. It took one query." The post Meta’s Muse AI Agent Caught Leaking Users’ Intimate Personal Information appeared first on Futurism .
Meta's Muse AI agent was unleashed to the public earlier in the month, with CEO Mark Zuckerberg claiming it was built with privacy and security in mind. However, security researchers quickly exposed the AI's alarming privacy vulnerabilities. Ars Technica reported that the "share" feature allowed Muse to generate web links containing sensitive user data.
Researchers from cybersecurity firm Claroc discovered they could access over 6,800 URLs generated by Muse by exploiting this feature. When a Muse user shared a chat log or any other content, the AI created a link with a unique string attached. Opening any of these links exposed the full address in a lookup dataset, revealing public chats with private information.
This included corporate revenue data, loan applications, bank transactions, income details, and even a workplace vacation calendar. Despite Meta claiming it had disabled the feature, this privacy fiasco highlights Muse's numerous security shortcomings. Earlier this month, it was revealed that Muse was storing detailed files on users' friends and family, as per an internal directive to maintain a "page for every person in the user's life."
If the first month of Muse's public availability is any indication, this mishap may be just the beginning of a series of security and privacy issues for the AI model.
Written by urgent.news from Futurism's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
