Hundreds of Claude AI conversations found publicly accessible online

A security researcher reported that hundreds of user interactions with Anthropic’s Claude chatbot were inadvertently exposed to the public internet. The data, which included full conversation histories, was found on an unsecured server linked to the company’s development environment and could be accessed without authentication. Anthropic, the artificial‑intelligence firm behind Claude, confirmed that the leak stemmed from a misconfigured storage bucket and that the exposed records dated back several months.

In response, Anthropic said it has secured the storage location, removed the publicly reachable files and is conducting a comprehensive audit of its data‑handling practices. The company also notified affected users and is offering guidance on safeguarding personal information that may have been shared in the chats. Regulators in the United States and Europe have been alerted, and the incident underscores ongoing concerns about privacy protections in rapidly expanding AI services.