HISPI Project Cerebellum HISPI PROJECT CEREBELLUM

Anthropic Said DeepSeek, Moonshot, and MiniMax Used Fraudulent Accounts and Proxies to Illicitly Distill Claude Capabilities at Scale

February 23, 2026

Anthropic discovered widespread activities involving fraudulent accounts and proxy services, which were used to generate extensive interactions with its AI model, Claude. These actions, attributed to DeepSeek, Moonshot, and MiniMax, aimed at distilling capabilities of the model for competitor training, potentially violating Anthropic's terms and access restrictions. The unauthorized activities resulted in millions of exchanges across thousands of accounts. Anthropic shared details about their detection measures, account controls, and indicator-sharing as responses.

Join us to learn more about responsible AI governance, incident mapping, and HISPI Project Cerebellum TAIM's role in measuring and managing such incidents.
JOIN US

Matched TAIM controls

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

Alleged deployer
deepseek, moonshot-ai, minimax, proxy-reseller-services
Alleged developer
anthropic
Alleged harmed parties
anthropic, claude-users, anthropic-customers, national-security-and-intelligence-stakeholders

Source

Data from the AI Incident Database (AIID). Cite this incident: https://incidentdatabase.ai/cite/1395

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.