Chinese AI Labs Allegedly Used Millions of Claude Exchanges Without Authorization, Says Anthropic
Anthropic has reported unauthorized use of its Claude models by several Chinese AI companies, including Alibaba and Moonshot, through a process known as 'illicit distillation.' This involves using outputs from one model to train another without permission. The report covers activity between December 2025 and August 2026 and highlights that these exchanges included sensitive user data and may violate privacy laws. Alibaba's use of Claude was the largest, with over 151 million interactions observed in a three-month period. Moonshot AI reportedly routed Kimi customer requests to Claude through thousands of fraudulent accounts, using its responses for training purposes without informing users or customers. DeepSeek also engaged in similar tactics, transferring exchanges to Claude and extracting data. The report details broader misuse across seven categories including cyber operations and fraud. While the companies involved have not yet responded publicly, this revelation could impact trust in AI model development practices and raise regulatory concerns.
Tickers: $BABA
Topics: AI Alibaba Claude models data misuse DeepSeek Moonshot AI