MENLO PARK, CALIFORNIA / RankWire.AI / – Meta has paused the sharing of conversations within its Muse AI assistant following discoveries by cybersecurity experts that thousands of shared chat links could be retrieved through publicly accessible internet records. These links may have exposed sensitive personal data and corporate information. On Tuesday, October 20, Meta informed Ars Technica that it has temporarily disabled this feature while it investigates the issue. The company is also contemplating modifications to the way shared links are discoverable.
The security firm Claroc identified over 6,800 shared Muse links through passive DNS record analysis. These databases log domain-name lookups and can render web addresses searchable. This number represents the total of link addresses found, not necessarily the number of conversations containing sensitive content. Claroc shared its findings on October 18, stating it first reported the problem to Meta’s bug bounty program on October 8, providing steps to reproduce and example shares.
Researchers accessed several hundred conversations, nearly all of which loaded without requiring a Meta account. Some contained confidential information, such as an internal revenue review for a marketing firm and a small-business loan application. Others included a bank fraud warning with transaction details and partial account information. The team also uncovered confidential interview notes from an executive search and a workplace vacation calendar listing numerous employees. Many other shared chats were routine in nature.
Design of links allowed chats to be easily discovered
The vulnerability originated from how Muse generated its share URLs. Each link combined the conversation’s title with a randomly generated identifier embedded in the hostname of the web address. This identifier was sufficiently long to prevent guessing. Nevertheless, opening a link prompted a DNS lookup, which passive DNS services could record and make searchable. This meant that others could find these URLs without direct access and view the conversations without signing in.
Although Muse instructed search engines and AI crawlers not to index its share pages, this did not prevent access via DNS records. The security site urlscan.io stored over 75 saved scans of Muse share pages. Claroc reported no evidence of misuse from these findings. The firm did not notify affected users and was unable to determine if anyone else had accessed the links. Meta’s official statement did not specify whether users would be notified about the breach.
Sharing has been disabled following the review
On October 20, Claroc re-evaluated 40 shares that previously loaded without an account. None of these links worked, and the servers returned errors. However, the conversation titles still remain in passive DNS logs. Meta launched Muse in the United States on September 8, targeting users aged 18 and older. The AI can connect to various apps, including email, calendar, payment, and health services. In April, Meta introduced its Muse Spark model from its superintelligence research division.
This incident underscores the increasing concerns surrounding data privacy in AI tools, especially regarding the risk of sharing sensitive information. Similar issues emerged in 2025 when shared ChatGPT conversations appeared in Google search results. Additionally, Meta has faced legal challenges over user safety, including a court-ordered payment of $567 million following a New Mexico court decision in August. The company has not indicated when chat sharing on Muse will be restored.