MENLO PARK, CALIFORNIA / RankWire.AI / – Following security researchers’ discovery of over 6,800 shared Muse chat links exposed through public DNS records, Meta has temporarily disabled the chat sharing feature within its Muse AI assistant. The breach could have potentially disclosed sensitive personal and corporate data. On Tuesday, October 20, Meta announced it had halted the feature while examining the security report. The company is considering modifications to prevent shared links from being easily discoverable.
The investigation was carried out by security firm Claroc, which found more than 6,800 Muse share links in passive DNS logs. These records track Domain Name System lookups, making website addresses accessible to anyone analyzing the data. Claroc notified Meta through its bug bounty program on October 8 and made its findings public on October 18. The count refers to discovered URLs, not the contents of conversations or sensitive information.
Researchers at Claroc examined a few hundred of the shared chats, and nearly all opened without requiring user login. Some examples included a loan application from a small business owner containing part of an account number, along with a note about a bank fraud alert. Additional examples featured a marketing firm’s revenue review prepared for its CEO and a real-estate investor’s report on mortgages and returns. Confidential interview preparation notes also appeared accessible.
Shared chat titles exposed through hostname design
The vulnerability stemmed from how Muse’s sharing URLs were constructed. Each link incorporated the conversation title and a random identifier within the hostname. While the identifier itself was not easily guessable, visiting a link triggered a DNS lookup that passive DNS services could log. This allowed anyone who never received the link to find it within these databases. When accessed, the shared conversations opened without requiring authentication.
Though Muse’s privacy settings instructed search engines and AI crawlers to ignore its share pages, researchers demonstrated that these pages could still be discovered without such tools. Over 75 saved instances of Muse share pages were stored on urlscan.io. Claroc confirmed it found no evidence of misuse of the data. It also stated that it did not contact the users affected and could not determine if others had uncovered the links. Meta’s official response did not specify whether users would be notified about the breach or future updates regarding the feature’s availability.
Temporary suspension amid security review
On October 20, when Claroc reexamined 40 of the previously accessible shared links, none of them loaded, returning errors. The conversation titles, however, remained visible in passive DNS history. Muse was launched in the United States on September 8, targeting users aged 18 and over. It integrates with email, calendar, payment, and health applications. Meta previously unveiled its Muse Spark model, created by its superintelligence lab, in April.
This incident underscores ongoing concerns regarding privacy risks in AI-enabled tools and the hazards of sharing chats containing sensitive information. In 2025, conversations shared via ChatGPT became indexable in Google search results, prompting OpenAI to remove the feature that allowed such indexing. In August, Meta was also ordered to fund teen mental health programs in New Mexico. The company has yet to specify when Muse’s sharing functionality will be reinstated or how it plans to modify link security measures.