Meta suspends AI conversation sharing feature following privacy concerns

Meta has temporarily deactivated the sharing functionality for its Muse AI assistant after security analysts identified a significant privacy exposure. Researchers at the firm Claroc reported that more than 6,800 unique links to AI conversations were accessible via public DNS records, allowing unauthorized parties to view private user interactions.

The vulnerability stemmed from the platform’s link-sharing mechanism, which generated URLs for AI-generated artifacts. While these links were intended to be private, their presence in public records made them discoverable. Muse spokesperson Tracy Clayton confirmed the suspension of the feature, stating that the company is currently reviewing the reporting process and evaluating potential modifications to link security.

This incident highlights ongoing data management challenges for companies deploying large language models. As AI platforms increasingly integrate sharing features, the risk of accidental data exposure through indexed URLs poses a growing concern for both developers and users regarding the privacy of interactions.