
Meta Pauses AI Chat Sharing Feature Following Security Discovery
Meta has temporarily disabled a feature allowing users to share AI conversations after security researchers discovered thousands of private links were publicly accessible. The company is currently reviewing the vulnerability to ensure user privacy.
Meta has suspended its 'Muse' feature, which allows users to share specific AI-generated conversations via unique URLs. The decision follows a report from security firm Claroc, which identified that over 6,800 of these conversation links were discoverable through public DNS records. This discovery suggests that links intended to be private or limited to specific recipients were potentially accessible to anyone capable of locating them in public databases.
In response to the findings, a Meta spokesperson, Tracy Clayton, confirmed that the company has paused the sharing functionality. The company stated that it is currently reviewing the report to address the security concerns. While the feature was designed to allow users to share 'Muse artifacts'—the content generated during interactions with the AI assistant—the current implementation allowed those links to be indexed or found outside of the intended user-to-user sharing process. Meta has not yet provided a timeline for when the feature might be restored or what specific technical changes will be implemented to prevent future exposure of these links.
📡 Media Analysis
How each outlet framed the story — angles, word choices, and what they chose to push or ignore.
Reported the technical failure and the company's immediate response without sensationalism.
"paused this functionality"
✓ Only outlet to report: Reported the specific number of exposed links (6,800) found by researchers.
⚡ Where Sources Disagree
- ·There are currently no conflicting reports regarding the facts of the incident.
🔍 What Nobody's Reporting
- ·Lack of comment from the security firm Claroc regarding their methodology for finding the links.
- ·No information on whether any sensitive user data beyond the AI chat content was exposed.
📰 Sources
0 A-rated source(s) among 1 total. Lowest trust: The Verge (B)
