Vulnerability in Meta's Muse AI exposed user data

A major security flaw in Meta's Muse AI assistant accidentally exposed confidential user and enterprise data. The vulnerability was caused by the way sharing links were structured, with conversation titles embedded directly into the subdomain of the web address, making them easy to identify through passive DNS searches.

Sensitive information easily accessible

Thousands of links containing sensitive information, such as financial reports and personal communications, could be discovered effortlessly. Meta responded by disabling the sharing feature to investigate the issue, but experts have raised concerns about the implications of the breach.

Comparison to past privacy scandals

Experts have criticized the privacy breach as similar to the Cambridge Analytica scandal. Meta describes Muse as a secure, privacy-respecting AI assistant, but the recent flaw has raised serious doubts about its claims. Muse operates within a dedicated virtual machine that hosts both the assistant and user data, yet its design has led to major security issues.

Security experts warn of invasive data collection

Early user experiences with Muse have highlighted serious security concerns. The Future Society and researcher Karan Joshi have pointed out that Muse gathers extensive data from user photos and messages to build detailed profiles of their surroundings. This approach has been described as "particularly alarming" due to its intrusive nature.

Web links expose confidential data

A major vulnerability in Muse's architecture allows personal and professional confidential data to be exposed when users share conversations via web links. Unlike standard online services, which use random characters in URLs to obscure them, Muse integrates conversation titles directly into the subdomain. This makes the links easily identifiable in passive DNS systems.

Researchers uncover thousands of vulnerable links

Security company Claroc discovered 6,800 links created by Muse users through a single query. These links, referred to as "GhostShares," can be accessed by anyone, even those who were never sent the link. A review of these links revealed highly sensitive information, including internal reports, financial details, and personal communications.

Meta responds to security concerns

Meta acknowledged the vulnerability after being alerted by security researcher Asaf Aprozper in early October. The company paused the feature and stated it would reevaluate it. This incident comes amid growing concerns about Muse's capabilities, which include performing tasks like filling out forms and placing orders.

Meta's history of privacy issues

Security experts have warned users not to trust Meta's assurances about privacy, citing its past issues such as the Cambridge Analytica scandal. Muse requires extensive access to users' personal information, including passwords and financial data, raising concerns about its impact on privacy and mental well-being.

Muse's potential for misuse

Journalist Eli Tan described using Muse as unsettling, likening it to giving away one's cognitive capacity. Author Ariana Bindman compared Muse to a "digital panopticon disguised as a fairy," emphasizing the company's focus on marketing over user privacy.

Zero-day vulnerability discovered

Security researcher Patrick Wardle found a zero-day vulnerability in Muse that allowed attackers to bypass permissions and access sensitive data and hardware without user consent. Meta fixed this flaw in late September 2026, but concerns about the application's security remain high.

Call for user caution

Experts urge users to be cautious about sharing personal data with Meta. They question the trustworthiness of the company's leadership, particularly regarding privacy and data handling. The ongoing security issues with Muse have raised serious doubts about its reliability and the safety of its users' information.