Brands
Claude chat links exposed in Google Search, raising fresh concerns over AI privacy
Anthropic fixes indexing issue after shared conversations revealed sensitive user information
New York: Some AI chats found more than an audience. A number of seemingly private conversations with Anthropic’s Claude chatbot briefly appeared in Google Search results, exposing sensitive information including cryptocurrency wallet keys and personal details after users unknowingly shared public links.
According to a report by Fortune, the issue stemmed from Claude’s “share” feature, which lets users generate a public URL to share conversations. Members of a Reddit discussion forum discovered that specific Google search queries surfaced lists of shared Claude conversations and Claude Artifacts, raising concerns about the privacy of information users believed was confidential.
Anthropic said it had not intentionally made chats searchable and attributed the issue to the way shared links were handled.
“We give people control over sharing their Claude conversations publicly, and in keeping with our privacy principles, we do not share chat directories or sitemaps with search engines like Google,” an Anthropic spokesperson told Fortune.
The company added that conversation links are accessible only when users choose to share them. However, once such links are created, they can be indexed or archived by third-party services, making them discoverable through search engines.
Anthropic has since addressed the issue by removing the indexed links from Google Search. However, previously generated public URLs remain accessible to anyone who already has those links.
The incident echoes similar privacy lapses across the AI industry. Last year, nearly 100,000 shared ChatGPT conversations were exposed through a comparable issue, while Grok, developed by xAI, also faced a similar vulnerability. In each case, publicly shareable URLs created by chatbot platforms were indexed by search engines.
Privacy experts warn that chatbot conversations often contain highly sensitive information because users increasingly rely on AI assistants to discuss work projects, finances, health concerns and legal matters. The latest incident highlights the continuing challenge for AI companies to balance convenient sharing features with robust privacy safeguards as chatbot adoption continues to grow.
