Hundreds of confidential user conversations and interactive projects from Anthropic's Claude AI chatbot appeared publicly accessible in search engine results, including Google.
The exposed data contained sensitive corporate project details, employee reviews, internal technical code, and highly confidential medical records with patient names.
How the Exposure Happened
The security oversight stemmed from Claude's shared chat feature, which generates accessible web addresses for users who want to distribute conversations.
Anthropic stated that users maintain direct administrative authority over whether their conversations are shared publicly across external platforms.
"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," said Amie Rotherham, Spokeswoman at Anthropic.
The company emphasized that individual shared links remain undiscoverable unless explicitly published online by users.
Remediation and Industry Context
Search engine indexing for the exposed chat links was subsequently blocked and remediated.
Search engines clarified that indexing depends directly on website directive controls and public web availability configured by site operators.
"We give site owners clear controls to decide whether pages can be crawled or indexed, and we always respect those directives," said Ned Adriance, Spokesperson at Google.
Similar search indexing incidents previously impacted competing AI platforms, including OpenAI's ChatGPT and xAI's Grok chatbot.
