New Delhi: AI startup Anthropic has patched a privacy flaw after public links of user conversations with its Claude AI chatbot briefly surfaced on Google Search. The issue primarily impacted users who actively used the chatbot‘s “share” feature, inadvertently exposing sensitive credentials, including cryptocurrency wallet keys.
The incident came to light after users flagged the exposure on Reddit, demonstrating how specific search queries on Google yielded indexable links to shared Claude conversations.
Shared Links vs Private Chats
Clarifying the situation, an Anthropic spokesperson told Fortune that user chats remain strictly private by default. The company stressed that it does not submit chat directories or sitemaps to search engines.
“We give people control over whether to share their Claude conversations publicly — and in keeping with our privacy principles, we do not share chat directories or sitemaps with search engines like Google,” the spokesperson stated. “These shareable links are not guessable or discoverable unless people choose to share them themselves. When someone shares a conversation, they are making that content publicly accessible, and like other public web content, it may be archived by third-party services.”
When a user taps the “share” button in Claude, the platform generates a standalone web snapshot of that specific chat intended for limited sharing. Anthropic confirmed these links do not grant access to a user’s wider account or unshared private conversations.
Issue Fixed, But Already-Exposed Links Persist
Anthropic has since addressed the indexing mechanism, removing the links from Google search results. However, reports indicate that while search engines no longer list these URLs, individual links that were previously exposed remain accessible to anyone who already possesses them.
This is not the first time AI platforms have faced search engine indexing troubles. According to Forbes, Claude experienced a similar exposure last year. OpenAI’s ChatGPT suffered a nearly identical vulnerability that exposed almost 100,000 shared conversations on Google, while Elon Musk’s Grok chatbot has also encountered comparable issues in the past.
