MENLO PARK, CALIFORNIA / RankWire.AI / – Meta has temporarily disabled the sharing of chat links in its Muse AI assistant following the discovery by cybersecurity experts that thousands of shared links could be located via public internet records. These links had the potential to reveal sensitive personal and corporate data. Meta informed Ars Technica on Tuesday, October 20, that the feature was put on hold while it assesses the findings. The company is also exploring modifications to how shared links are accessed and discovered.
Security firm Claroc identified over 6,800 shared Muse links by analyzing passive DNS records—databases that log domain-name lookups and make web addresses searchable. This count reflects the number of link addresses found, not the total number of sensitive conversations. Claroc shared its findings on October 18 and stated it first notified Meta’s bug bounty program on October 8, providing reproduction steps and sample shares.
Multiple conversations were accessed, with nearly all live shares opening without the need for a Meta account. Some of these contained confidential information, including an internal revenue review from a marketing firm and a small-business loan application. Others displayed a bank fraud alert with transaction details and partial account information. Researchers also uncovered confidential interview notes for an executive search and a workplace vacation calendar listing dozens of employees. Many other shared chats contained routine or non-sensitive material.
Design of share links made chats discoverable
The vulnerability originated from how Muse constructed its share URLs. Each link integrated the chat’s title with a randomly generated identifier within the hostname of the web address. This identifier was sufficiently lengthy to prevent guessing. Nonetheless, opening a link prompted a DNS lookup, which passive DNS services could record and make searchable. This allowed others to find these links without directly receiving them, and view conversations without logging into a Meta account.
Although Muse instructed search engines and AI crawlers not to index its shared pages, this did not prevent access via DNS records. The security site urlscan.io stored more than 75 saved scans of Muse share pages. Claroc indicated it found no signs of misuse, and it did not contact affected users nor determine if others had discovered the links. Meta’s official statement did not specify whether users affected would be notified.
Sharing capabilities now turned off following review
On October 20, Claroc reverified 40 links that previously opened without an account, and all failed to load, with the content-serving addresses returning errors. However, the conversation titles still remain in passive DNS history. Meta launched Muse in the U.S. on September 8, targeting users aged 18 and older. The AI assistant can integrate with email, calendar, payment, and health apps. In April, Meta introduced its Muse Spark model developed by its superintelligence research division.
This incident underscores increasing worries about data privacy in AI tools and the dangers of sharing conversations containing sensitive details. Similar issues arose in 2025 when shared ChatGPT conversations appeared in Google search results. Meta has faced legal challenges related to user safety, including an order to pay $567 million following a New Mexico court ruling in August. The company has not provided a timeline for when Muse chat sharing might resume.