THE CRUNCH
Meta has paused a Muse feature that lets people share their AI conversations through web links, after security researchers showed the links could be found and opened by anyone. Researchers at the security firm Claroc say they enumerated more than 6,800 share links Muse users had created, and after opening a few hundred found material clearly not meant for public eyes, including 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.
The problem lies in how Muse builds its share addresses. According to Claroc, each link is constructed from the share's title, a short phrase describing the conversation and a random ID. Because the hostname usually revealed what a page contained before anyone opened it, and the link carries the same title and ID, a hostname pulled from public DNS records, which act as a directory for web traffic, was enough to rebuild the link. That defeats the usual protection for share pages, where a long unguessable string in the URL keeps unintended people out.
Muse spokesperson Tracy Clayton said in an emailed statement that people can share specific Muse artifacts, creating a page accessible to anyone with the link, but that the company has paused the functionality while it reviews the report and considers changes to how the links can be discovered. The episode follows a familiar pattern: conversations shared from OpenAI's ChatGPT, Anthropic's Claude and xAI's Grok have previously ended up visible in Google Search.
IN NUMBERS