MENLO PARK, CALIFORNIA / RankWire.AI / – Meta has temporarily halted the chat link sharing feature in its Muse AI assistant after security researchers revealed that thousands of shared conversation links could be accessed via public internet records. This exposure risked revealing sensitive personal and corporate information. On Tuesday, October 20, Meta announced it had paused the feature while reviewing the incident report. The company is considering modifications to how shared links are discoverable.
The discovery was made by security firm Claroc, which identified more than 6,800 Muse share links in passive DNS data. These records log Domain Name System lookups that can make website addresses accessible to anyone who searches the data. Claroc stated it informed Meta through its bug bounty program on October 8 and published its findings on October 18. The 6,800 count reflects the number of discovered addresses, not sensitive conversation content.
A few hundred of these conversations were reviewed by Claroc’s researchers, who found nearly all shared links accessible without requiring login credentials. Among the exposed data were a small-business loan application with a partial account number and a note about a bank fraud alert. Additional content included a marketing firm’s revenue report prepared for its CEO and a real-estate investor’s report on mortgages and returns. Confidential interview notes were also visible among the shared data.
Hostnames exposed chat titles
The root of the issue was in how Muse’s sharing addresses were constructed. Each link embedded the conversation title along with a random identifier within the hostname. While the identifier was not easily guessable, accessing a link prompted a DNS lookup that passive DNS services could record. Individuals who never received the link could then locate it within these databases. Importantly, the shared conversations opened without requiring user authentication.
Muse’s configuration advised search engines and AI crawlers to skip its sharing pages, yet researchers did not need a search engine to locate them. Over 75 saved scans of Muse share pages appeared on urlscan.io. Claroc clarified that there was no evidence of data misuse and did not reach out to affected users, also noting it could not determine if others had discovered the links. Meta’s statement did not specify whether users would be informed of the incident.
Sharing feature put on hold during review
On October 20, when Claroc re-examined 40 previously accessible shares, none opened, and their addresses returned errors. However, the conversation titles still appeared in passive DNS history. Muse launched in the United States on September 8, targeting adults aged 18 and over. The platform can connect to users’ email, calendar, payment, and health applications. Earlier, Meta introduced its Muse Spark model, created by its superintelligence lab, in April.
This incident adds to ongoing concerns regarding privacy within AI tools and the potential risks associated with sharing conversations containing confidential information. In 2025, shared ChatGPT chats appeared in Google search results, prompting OpenAI to remove the feature that allowed indexing. Additionally, in August, Meta was ordered to fund teen mental health programs in New Mexico. Meta has yet to announce when Muse’s sharing feature will be restored or how the link-sharing process might be altered.