China's Cyberspace Administration probes DeepSeek and Moonshot over alleged data leaks to Anthropic via Claude, staff questioned
anthropic claude deepseek
| Source: Techmeme | Original article
China's internet regulator is probing DeepSeek and Moonshot over alleged data leaks to Anthropic via Claude, and has questioned staff.
China’s Cyberspace Administration has opened an investigation into two domestic AI firms, DeepSeek and Moonshot AI, after allegations surfaced that they silently routed user queries to Anthropic’s Claude model and exposed sensitive data in the process. According to The Information, regulators have summoned staff from both companies for questioning. The probe follows reports that a user linked to the People’s Liberation Army inadvertently fed surveillance footage from hundreds of cameras in Chengdu—including feeds near PLA installations—into Claude, thereby leaking military‑grade data to a U.S.‑based service.
Anthropic has publicly accused DeepSeek and Moonshot of “secretly routing” more than 300,000 user requests to Claude and then re‑branding the responses as their own. The claim, echoed by several Chinese media outlets, suggests a covert data‑flow pipeline that could give the Chinese firms access to advanced generative‑AI capabilities while simultaneously compromising Chinese user privacy and national‑security information.
The investigation matters on several fronts. First, it highlights the growing friction between China’s push for home‑grown AI leadership and the reliance on foreign models to accelerate development. Second, it raises questions about data‑sovereignty and the enforcement of China’s strict cybersecurity laws, especially where foreign cloud services are involved. Finally, the episode could intensify scrutiny of cross‑border AI collaborations, a sector already under pressure after recent reports of OpenAI negotiating stress‑test agreements with Anthropic.
What to watch next: whether the Cyberspace Administration, the Ministry of Industry and Information Technology, the Ministry of Public Security or the Ministry of National Defense will issue formal findings or penalties; how DeepSeek and Moonshot respond to the allegations; and whether Anthropic or other U.S. AI firms will adjust data‑routing practices to avoid similar disputes. The outcome could set a precedent for how China regulates AI models that incorporate foreign‑origin components.
Sources
Back to AIPULSEN