Anthropic Accuses Moonshot, DeepSeek of Data Misuse and Fraudulent AI Practices
September 12, 2026
Anthropic accuses Moonshot and DeepSeek of covertly routing user requests to Claude and presenting Claude outputs as their own, raising concerns about US customer data exposure.
Moonshot is positioning itself as a leading China AI player, with its K3 model released in July and projections from Nomura Securities that ARR could reach about US$1 billion by year-end.
Moonshot allegedly diverted roughly 300,000 requests to Claude over ten days using 5,380 fraudulent accounts, with most accounts traced to Singapore and Japan; details on DeepSeek’s scale were not disclosed.
Anthropic characterizes the practice as illicit distillation on an industrial scale, enabled by fake accounts and stolen payment methods, potentially violating US and Chinese laws.
Other Chinese firms, including Alibaba’s Qwen, Z.ai (MiMo), Xiaomi’s M model, and MiniMax, are also accused of distilling Claude outputs, though with varying levels of detail.
Anthropic’s 154-page report claims Moonshot accessed data from Moonshot’s US, European, and Chinese users, including a user possibly linked to China’s People’s Liberation Army, as well as data from other high-profile Chinese tech firms.
Summary based on 1 source
Get a daily email with more AI stories
Source

The Star • Sep 12, 2026
Moonshot, DeepSeek secretly routed user requests to Claude, Anthropic claims