MENLO PARK, CALIFORNIA / RankWire.AI / – Meta has paused the sharing of conversation links within its Muse AI assistant after cybersecurity experts identified that thousands of shared chat URLs could be located via publicly accessible internet records. These links risked exposing private personal data and corporate details. On Tuesday, October 20, Meta informed Ars Technica that it had temporarily disabled the feature as it investigates the security issue. The company is also exploring potential modifications to prevent shared links from being easily discoverable.
Security firm Claroc detected over 6,800 shared Muse links through passive DNS records, which track domain name lookups and allow searchability of web addresses. This number represents discovered link addresses rather than the count of sensitive conversations. Claroc released its findings on October 18, reporting that it had first submitted the issue to Meta’s bug bounty program on October 8, including reproduction steps and sample shares.
The researchers examined several hundred conversations, and nearly all shared links accessible without a Meta account. Some of these chats contained sensitive material such as an internal revenue review for a marketing firm and a small-business loan application. Others included a bank fraud alert with transaction figures and partial account information. Additional shared content included confidential interview notes for an executive search and a workplace vacation calendar displaying numerous employees. Many other chats involved routine discussions.
Design of Share Links Made Conversations Discoverable
The vulnerability originated from the way Muse generated its share URLs. Each link combined the conversation’s title with a randomly created identifier embedded in the hostname of the URL. While the identifier was sufficiently long to prevent guessing, opening a link prompted a DNS lookup that could be recorded by passive DNS services, rendering the URLs searchable. This meant that others could locate the links without direct access and view the chats without logging in.
Although Muse instructed search engines and AI crawlers not to index its shared pages, this did not prevent DNS-based access. The security site urlscan.io maintained over 75 stored scans of Muse share pages. Claroc stated it found no evidence of misuse. The firm did not notify affected users and was unable to determine if others had discovered the links. Meta’s official statement did not clarify whether users impacted by this issue would be informed.
Sharing of Links Suspended Following Internal Review
On October 20, Claroc re-examined 40 links that previously loaded without a Meta account. None of these now functioned, and the servers hosting the content returned errors. However, the conversation titles remain visible in passive DNS records. Meta launched Muse in the United States on September 8, targeting users aged 18 and older. The assistant is capable of integrating with email, calendar, payment, and health applications. In April, Meta introduced its Muse Spark model, developed in its superintelligence laboratory.
This incident underscores increasing worries about data privacy in AI applications, especially when sharing conversations that may include confidential or sensitive information. Similar issues emerged in 2025 when shared ChatGPT discussions appeared in Google search results. Meta has also faced legal scrutiny regarding user safety, including a ruling requiring the company to pay $567 million following a New Mexico court judgment in August. Meta has not announced when Muse’s sharing feature might be reinstated.