Hundreds of conversations users had with Claude, Anthropic’s AI chatbot, were found to be available online for public viewing.

The issue was discovered by a Reddit user who used a specific Google search that used specific keywords bringing up specific user chats with Claude.

Claude chats that showed up in the Google Search were those where a user had clicked to “share” a link, which had been saved by the search engine, meaning they were then observable to the public. Chats where this share link was clicked also appeared on other search engines such as Duck Duck Go, Bing, and Brave.

The “share chat” option on Claude allows users to create a link to specific chats that they can share with others. The interface says that “Anyone with the link can view” similar to the share option available on Google Docs.

How did these chats end up available on search engines then?

A spokesperson for Anthropic said that only chats shared publicly and where Search Engines could see – for instance, social media platforms or online forums – while a privately shared link would not be accessible on a search engine, according to TechCrunch.

Amie Rotherman, an Anthropic spokesperson, explained: “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.

“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.”

Recommended reading

• Claude Source Code Leak Reveals Anthropic’s Secret Plans
• Claude Goes Quiet in System Outage, Users Confirm
• Staffers Resist Change as Anthropic Sues US Gov Over AI Contract Quagmire
• OpenAI Backtracks on “Opportunistic and Sloppy” US Government Deal

The issue was first reported on Reddit Saturday, but appears to have been disabled as the same search method does not bring up Clause user conversations.

However, many user conversations were seen across the internet, some of which including personal information like names, work places, and medical information.

Claude conversations that generated erotica were also found on the internet, despite Anthropic guardrails that supposedly prevent Claude from generating sexually-explicit material.

Anthropic has already faced similar issues in the past – hundreds of Claude chats were found on search engines last year as reported by Forbes, which were then taken down by Anthropic.

Other AI chatbots have experienced the same, with ChatGPT and Grok conversations found via search engines.