A security firm discovered that user conversations were indexing in public web archives without any required authentication. Security firm Claroc identified more than 6,800 shared conversations from Meta’s Muse AI chatbot in a DNS database, which were accessible via web links that anyone could open without a password. The exposed data included sensitive files such as bank fraud alerts, loan applications and confidential company documents.
Meta has suspended the sharing feature of its chatbot to address the vulnerability, although records of the leaked links remain publicly accessible. The breach potentially exposes people who never used the service if their private information was discussed in a chat shared by another user, creating risks of identity theft and unauthorized data profiling.
