Chinese regulator probes DeepSeek and Moonshot over secret data routing
China is investigating DeepSeek and Moonshot for alleged sensitive data leaks to Anthropic
Chinese regulators are investigating DeepSeek and Moonshot AI over concerns regarding how both companies route and handle sensitive user data to Claude models, raising questions about data protection and information security, according to a report from The Information.
The report surfacing on September 10 outlined how seven Chinese companies were using Claude via mass unauthorized access. It alleged that DeepSeek escalated engineering requests involving law enforcement monitoring infrastructure to Claude.
The Cyberspace Administration of China summoned delegates from the seven companies named in the report for several rounds of questioning following its release. Meanwhile, other five companies identified were Alibaba, Zhipu, SenseTime, MiniMax and Xiaomi.
This report marked the first example where the US AI firm detailed substantial data to support such claims. Moonshot primarily sent over 23 million interactions from May to July while DeepSeek handled more than 12 million in just 14 days in July which involved data related to People’s Liberation Army facilities, police data and corporate secrets.
In addition, officials are evaluating the violations amid Beijing’s strict rules on cross-border data and the ongoing US-Chinese AI rivalry though no sanctions have been announced.
