Moonshot, DeepSeek secretly routed user requests to Claude, Anthropic claims

American artificial intelligence (AI) giant Anthropic has accused its Chinese AI rivals DeepSeek and Moonshot on Thursday of covertly routing customer requests to its Claude models, then showing users the responses as if they were from their own models.

The new allegations mark a further escalation in US criticism of China’s AI industry practices. Earlier this week, three US government agencies, including the National Security Agency (NSA) and the Federal Bureau of Investigation (FBI), claimed that six Chinese companies were engaged in an industrial campaign to steal US model outputs, “likely with Chinese government awareness”.

Beijing on Wednesday rejected those allegations, describing them as actions by Washington to suppress China’s AI industry and threatening retaliatory measures.

While US companies OpenAI and Anthropic have repeatedly accused their Chinese counterparts of “stealing” their outputs to improve their own models in a process known as “distillation”, Anthropic’s latest claims mark the first time it has accused Chinese companies of re-routing users directly to Claude, which would expose Chinese users’ information and activity to the American company.

“[These] findings raise concerns about the misuse of user data by PRC AI labs,” the company said in its 154-page report released on Thursday.

25:46

Post Techcast: the great AI schism

According to the report, Kimi model developer Moonshot secretly showed its users Claude model outputs and passed them off as Kimi’s, then used those exchanges to improve their own models.