Hundreds of conversations from Anthropic's AI chatbot Claude briefly appeared in Google Search results after users discovered that publicly shared chat links had been indexed by search engines. The exposed conversations reportedly included discussions containing personal information, health-related queries, legal advice and cryptocurrency details.
The issue came to light after Reddit users found that searching "site:claude.ai/share" on Google returned publicly accessible Claude conversations. The indexed pages included both chat histories and Claude Artifacts, interactive documents and applications created within the chatbot.
According to reports, the searchable conversations covered a range of topics, including programming, work-related discussions and personal matters. Wired reported that the indexed pages remained visible in search results for at least two days before they were removed.
Anthropic said the incident stemmed from its chat-sharing function, which creates public links that users can distribute to others. In a statement, the company noted that it does not supply chat directories or sitemaps to search engines, and that these shared links remain undiscoverable to the public unless users actively choose to share them.
"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 said.
The company further noted that once a conversation is made public, it may be indexed by search engines just like any other publicly available webpage.Google said search engines index publicly accessible webpages and that website owners can prevent indexing using tools such as robots.txt and noindex tags.
Similar incidents have previously affected ChatGPT and Grok, renewing concerns over how AI chat-sharing features can expose sensitive conversations once they become publicly accessible.