Claude AI Shared Chats Briefly Indexed by Google and Bing, Raising Fresh Privacy Concerns
Anthropic is facing renewed scrutiny after publicly shared Claude AI conversations briefly became searchable through Google and Bing, exposing sensitive user discussions to anyone who knew how to find them. Although the indexed pages have largely been removed, the incident highlights the privacy risks of sharing AI conversations through public links.
The exposure affected shared conversations, not users’ private chat histories.
Shared Links Became Searchable
The issue was first identified by Reddit users, who discovered that Claude conversations shared through Anthropic’s public link feature were being indexed by search engines.
Using searches that included site:claude.ai/share, users were able to locate thousands of publicly accessible conversations and shared Claude artifacts.
While the feature is intended to let users share conversations with specific people via a link, those links were apparently accessible to web crawlers once published elsewhere online.
Sensitive Information Was Exposed
Among the conversations reportedly indexed were discussions containing:
- Medical records
- Personal information
- Internal corporate data
- Explicit content
Shared Claude Artifacts—including applications, visualizations, and other AI-generated content—were also discoverable through search engines.
Most of the indexed pages have since been removed from Google and Bing search results.
Not the First Time
This is not the first incident involving searchable Claude conversations.
A similar situation was reported in 2025, when hundreds of publicly shared Claude chats appeared in Google Search before later being removed.
The latest incident has renewed concerns about how public sharing features interact with search engine indexing as AI tools become more integrated into users’ daily workflows.
Anthropic Says the System Worked as Intended
Anthropic maintains that the exposure did not result from a security breach.
According to the company, users who generate a public share link are intentionally making that conversation publicly accessible.
An Anthropic spokesperson stated that:
- Shared links are not easily guessable.
- Conversations become discoverable only if users share those links publicly.
- Like any publicly accessible webpage, shared chats may be indexed or archived by third-party services.
Review Your Shared Conversations
Users who have previously shared Claude chats can review and revoke access through the application’s privacy settings.
To manage shared conversations:
- Open Settings.
- Navigate to Privacy.
- Select Shared Chats.
- Review existing public links and remove any that should no longer be publicly accessible.
A Reminder About AI Privacy
As AI assistants increasingly handle personal, financial, legal, and business-related tasks, this incident underscores the importance of treating public share links as truly public.
Before sharing any AI conversation, users should ensure it contains no confidential information, passwords, financial details, medical records, or other sensitive data. Even when a link is intended for a limited audience, it may become accessible far more broadly if shared publicly and subsequently indexed by search engines.
