Claude AI Chats Indexed by Google, Exposing User Conversations

Executive Briefing

• Reveals that Claude share links were searchable on Google, exposing legal, technical, and personal conversations publicly.
• Confirms Anthropic patched the vulnerability after a viral Reddit post highlighted the 'site:claude.ai/share' search method.
• Notes this marks the second such incident for Claude users, with a similar exposure occurring in September.
• Warns that AI chats are broadly not private, as Anthropic retains deleted conversation data for 30 days.
• Cautions users that most AI companies use chat data for model training unless users explicitly opt out.