Thousands of seemingly private conversations with Anthropic’s Claude AI chatbot were indexed by Google last weekend. Some contained sensitive information such as cryptocurrency wallet keys, names, and addresses. The issue affected conversations shared through Claude’s share feature, which generates a link meant to be sent to one person. Anthropic has fixed the flaw, but many links remained accessible. This is not the first such incident in the AI industry: nearly 100,000 ChatGPT conversations were similarly exposed last year.
Source: Read the original article
