Written by Lucy Hall and reviewed, fact-checked and signed off by a SocialDay editor before publication. Read our editorial standards and corrections policy. Spotted something wrong? Tell the newsroom.
Researchers found shared chat links by searching public DNS logs in a single query. The flaw exposes how AI sharing features can leak what you're discussing before anyone opens the link.
Meta has suspended the conversation-sharing feature in its Muse AI agent after security firm Claroc discovered that all shared Muse chat links could be found and accessed via a passive DNS search because every URL used the same domain name and included the conversation's title in the subdomain.
Claroc identified 6,800 links for shared Muse chats and accessed conversations containing sensitive information. The material included meeting details involving a leading AI lab, a small-business loan application, a bank fraud alert with partial account information, a company's vacation calendar and internal business documents.
The problem wasn't a breach. The exposure involved a sharing feature that worked as designed: when users asked Muse to share conversations, the assistant created public pages accessible to anyone holding the link. Researchers discovered that outsiders could find those links through passive DNS records without receiving them.
com domain, and because the conversation title appeared in the hostname, the links could be identified through passive DNS searches, which preserve historical records of domain activity.
Source link
