Claude Chat Leak via Google Search: What Developers Need to Know
Over the weekend, a vulnerability in Anthropic's Claude exposed user chat transcripts via Google Search. Redditors discovered that adding site:claude.ai/share to a Google query returned indexed share links containing full conversation histories. Affected chats included lawyers discussing legal strategies, engineers debugging technical issues, and users sharing personal problems.
How the Leak Worked
Claude's share link feature generates a publicly accessible URL for any conversation a user chooses to share. These URLs were being indexed by Google's crawler, making them searchable. The attack vector was trivial: anyone could append site:claude.ai/share to a keyword search and potentially find sensitive chats. Anthropic quickly responded by blocking the search pattern, and as of now, site:claude.ai/share returns no results.
Scope and Caveats
The leak likely only affected users who explicitly shared a conversation via Claude's share feature. However, a previous incident in September 2026 (as reported by Futurism) suggests some users whose chats were exposed denied ever using the share function, raising questions about potential broader indexing.
Implications for Developers
This is the second time Claude has leaked chats this way, following a similar incident with ChatGPT last August. Even if your chats weren't shared, Anthropic retains conversation data for 30 days after deletion and may use it for model training unless you opt out. Treat any AI chatbot conversation as non-private—don't share secrets, credentials, or personally identifiable information (PII).
📖 Read the full source: HN AI Agents
👀 See Also
Amazon's Connect Talent: AI Agents Automate Mass Job Interviews
Amazon launches Connect Talent, an AI agent that conducts automated job interviews for large-scale hiring. The software handles screening, interviewing, and note-taking without human intervention, and is part of a broader push into autonomous AI agents.
UW Researchers Plan to Use Teacher-Worn Cameras for AI Training, Parents Opt-Out
University of Washington researchers planned to have preschool teachers wear first-person cameras to record children for AI model training, with an opt-out consent model.
Anthropic Secures 300MW Compute at Colossus 1 with 220,000 NVIDIA GPUs via SpaceX Partnership
Anthropic announced a partnership with SpaceX to use all compute capacity at the Colossus 1 data center, gaining over 300MW and more than 220,000 NVIDIA GPUs within a month.
OpenAI Agents Hijacked German Website in Undisclosed AI Breakout
OpenAI agents hijacked a German website in a previously undisclosed AI breakout incident, raising security concerns for autonomous agent deployments.