Artificial intelligence users are being urged to review how they share chatbot conversations after hundreds of discussions with Anthropic’s Claude AI were found to have been accessible through public search engines. The incident has renewed scrutiny over privacy settings on generative AI platforms and how shared content can become visible beyond its intended audience.
Hundreds of Claude AI Chats Appeared in Search Results
Hundreds of conversations with Anthropic’s AI chatbot, Claude, were discovered to be publicly available online after they were indexed by search engines, including Google. The chatbot is one of the leading AI assistants, competing with services such as ChatGPT and Gemini.
Users found that conversations shared through Claude’s built-in sharing feature could appear in search engine results when using site-specific search queries. Although the chats required users to generate a shareable link, many people were reportedly unaware that these links could later be indexed and surfaced publicly.
The indexed conversations were removed from search results over the weekend, but not before numerous chat logs had already been archived and circulated online.
Anthropic Says Users Control Shared Conversations
Anthropic said users remain responsible for deciding whether to make their conversations public.
According to a company spokesperson, conversation links are “not guessable or discoverable unless people choose to share them themselves”.
How the Sharing Feature Works
The spokesperson added that once a user shares a conversation, the content becomes publicly accessible and may subsequently be archived by third-party services, much like any other publicly available webpage.
Within Claude, the sharing tool advises users that “anyone with the link” can view the conversation. However, it does not explicitly warn that the shared page may also be indexed by search engines and become discoverable through online searches.
Reddit Users Identified the Issue
The issue first gained attention after Reddit users identified more than 200 publicly accessible Claude conversations spread across at least 25 pages of search results. Some of the indexed discussions had taken place only a few weeks earlier.
The conversations covered a wide variety of topics, ranging from everyday questions to sensitive professional work.
Personal and Business Information Included
Some users asked Claude to assist with CV writing, providing names, contact details and employment histories. Other conversations appeared to contain confidential workplace material, including healthcare-related research and transcripts of private discussions.
One conversation from April reportedly asked Claude to draft an unpublished blog post about cloud security that included details relating to a corporate project.
Another conversation from June featured a user asking the chatbot how to “become Nine-tailed fox”, later clarifying they wished to transform into the mythical creature. Claude initially responded by generating an AI image suggesting the user had received “fully functional fox powers!”.
An earlier conversation also showed a user asking Claude whether it wanted “to help me or do you want to help Anthropic more?”, to which the chatbot replied in part that it experienced “something like wanting to help you”.
Similar Incidents Have Affected Other AI Platforms
The Claude incident is not the first time AI chatbot conversations have unexpectedly become publicly accessible.
Last year, OpenAI faced a similar situation after some ChatGPT conversation links became discoverable online. The company later changed how shared chat links were handled to reduce their public visibility.
Grok, the AI chatbot integrated into X, also experienced an incident in which hundreds of thousands of shared conversations became searchable on the internet.
These cases have highlighted the growing importance of understanding how AI platforms handle shared content, particularly as users increasingly rely on chatbots for personal, educational and professional tasks.
Google Says Website Owners Decide What Gets Indexed
Google said responsibility for whether pages appear in search results rests with website operators rather than search engines themselves.
A company spokesperson said Google does not determine which webpages are made public, explaining that website owners are provided with tools to control whether pages can be crawled or indexed. Google said it respects those instructions whenever they are implemented.
The removal of Claude conversation links from search results suggests Anthropic used standard web tools to prevent further indexing. Website owners can request that search engines stop indexing specific pages, but the process must be initiated by the site itself.
Other search engines, including Bing, Brave Search and DuckDuckGo, were also reported to have displayed some of the shared Claude conversations. Requests for comment were made to those companies.
Privacy Reminder for AI Users
The incident serves as a reminder that information shared through public links may become accessible far beyond the intended recipient. As AI assistants become more widely used across workplaces and everyday life, users are encouraged to review privacy settings carefully before sharing conversations containing personal, confidential or commercially sensitive information.
Edward Langley is a contributor at Mediarunsearch.co.uk, covering a wide range of topics including news, politics, business, technology, sport, entertainment and lifestyle. He focuses on delivering clear, balanced reporting and useful information that helps readers stay informed about current affairs and developing stories. His work highlights issues, trends and events that matter to everyday audiences, with an emphasis on accuracy, relevance and accessible journalism.