Meta Muse's share feature exposes sensitive user data through easily enumerable URLs
Researchers discovered that Meta's AI agent Muse allows attackers to discover and access shared conversations containing confidential business and personal information through a fundamental design flaw in its link-sharing mechanism.
What happened
Researchers from security firm Claroc reported Monday that Meta's Muse AI agent exposes sensitive information when users share conversations via shareable links. The researchers enumerated 6,800 links created by Muse users and found that opening a few hundred revealed confidential data not intended for public access, including executive-search notes, internal revenue reviews, loan applications, bank account details, real-estate performance reports with property addresses and financial figures, and employee vacation calendars. The vulnerability stems from how Muse constructs its share URLs: while the links contain long sequences meant to be unguessable, a central design decision makes them easily discoverable through passive DNS searches—a technique that requires only a single query. Claroc co-founder Asaf Aprozper reported the findings through Meta's bug bounty program on October 8 and attempted to contact Meta via LinkedIn without response. In response, Meta suspended the sharing feature while reviewing the issue and considering changes, stating that the company opened its Muse bug bounty program to help improve the product.
Context
Muse represents Meta CEO Mark Zuckerberg's stated commitment to building AI tools "from the ground up for privacy and security." The exposure of this vulnerability undermines that positioning and exposes users ranging from job candidates to business owners to workplace administrators to unintended disclosure of sensitive information. The flaw is particularly significant because it requires minimal technical effort to exploit—researchers characterized the attack as requiring only a single query—meaning malicious actors could systematically harvest confidential data at scale. The pause of the sharing feature affects users' ability to collaborate and share work through the platform while Meta determines how to redesign the URL structure to prevent enumeration.