HISPI Project Cerebellum HISPI PROJECT CEREBELLUM

Reported Public Exposure of Over 100,000 LLM Conversations via Share Links Indexed by Search Engines and Archived

July 31, 2025

In the years 2024 and 2025, sharing functions in various Language Learning Models (LLMs)—such as ChatGPT, Claude, Copilot, Qwen, Mistral, and Grok—reportedly led to unauthorized exposure of user conversations marked as 'discoverable' to search engines and archiving services. This incident resulted in approximately 100,000 chats being indexed, scraped, and potentially disclosed, allegedly exposing sensitive information such as API keys, access tokens, personal identifiers, and confidential business data. Such incidents underscore the importance of trustworthy AI, safe and secure AI practices, and the need for robust AI governance mechanisms like those provided by Project Cerebellum's AI Incident Database.

For those interested in shaping the future of AI and promoting harm prevention through guardrails for AI, JOIN US.

Matched TAIM controls

Suggested mapping from embedding similarity (not a formal assessment). Browse all TAIM controls

Alleged deployer
openai, anthropic, microsoft, xai, alibaba, mistral
Alleged developer
openai, anthropic, microsoft, xai, alibaba, mistral
Alleged harmed parties
users-of-chatgpt, users-of-claude, users-of-copilot, users-of-grok, users-of-qwen, users-of-mistral, general-public

Source

Data from the AI Incident Database (AIID). Cite this incident: https://incidentdatabase.ai/cite/1186

Data source

Incident data is from the AI Incident Database (AIID).

When citing the database as a whole, please use:

McGregor, S. (2021) Preventing Repeated Real World AI Failures by Cataloging Incidents: The AI Incident Database. In Proceedings of the Thirty-Third Annual Conference on Innovative Applications of Artificial Intelligence (IAAI-21). Virtual Conference.

Pre-print on arXiv · Database snapshots & citation guide

We use weekly snapshots of the AIID for stable reference. For the official suggested citation of a specific incident, use the “Cite this incident” link on each incident page.