Meta pauses Muse chat sharing after researchers find conversations on the open web

Meta has paused a feature in its Muse AI assistant that lets users share conversations, after researchers at the security firm Claroc found the shared pages publicly discoverable on the internet. Ars Technica first reported the issue, and the decision follows a report Claroc published this week.

How the links were found

Claroc says it located more than 6,800 Muse links in public DNS records, which work as a directory for web traffic. The feature creates a page that anyone with the link can open. However, the way the links were built made them easy to discover: Muse forms each web address from the title of the share, a short phrase describing the conversation, followed by a random ID. According to Claroc, the hostname often revealed what a page contained before it was opened, and a hostname taken from DNS could be used to rebuild the full link.

What was exposed

The researchers were able to view hundreds of Muse conversations. They included a marketing company's revenue review, a small business owner's loan application, a workplace's full-year staff calendar, and notes for a job candidate's call with an AI lab.

Meta's response

Muse spokesperson Tracy Clayton told The Verge in an emailed statement that users can share specific Muse artifacts, creating a page available to anyone who has the link. The company has paused the functionality while it reviews the report and considers changes to how the links can be discovered.

This is not the first time AI sharing features have made chats more public than intended. Conversations with OpenAI's ChatGPT, Anthropic's Claude and xAI's Grok have previously ended up in Google Search.