Claude Shared Chats Exposed in Google and Bing Search Results
View original source →On July 27, Wired and TechCrunch revealed that user-created public Claude share links were being indexed by Google and Bing, making sensitive conversations discoverable to anyone running a search. Anthropic fixed the Google exposure by end of day; Bing still showed approximately 612 results at time of publication.
Key Points:
• Searching site:claude.ai/share or site:claude.ai/public/artifacts surfaced publicly shared conversations in search results • Exposed content included API keys, cryptocurrency wallet information, personal resumes with home addresses, internal company project details, and legal discussions • Root cause: Anthropic used robots.txt to prevent indexing — but both Google and Bing explicitly document that robots.txt alone is insufficient; noindex tags or X-Robots-Tag headers are required • Anthropic remediated Google exposure within hours; Bing remediation was ongoing • This follows a similar ChatGPT incident in 2025 that OpenAI resolved by removing the sharing feature entirely
The gap between 'I shared this with a link' and 'this is now indexed and searchable' is the most common source of AI privacy failures. As AI products scale to hundreds of millions of users with sharing features, privacy architecture failures become a mass-harm risk.
Why It Matters: This exposure demonstrates that AI chats are now a category of sensitive user data requiring the same privacy controls as email or cloud storage. Using robots.txt alone for privacy is a documented technical error.