China Probes DeepSeek and Moonshot Over Secret Data Routing to Claude

China's Cyberspace Administration opened a formal data-security probe into DeepSeek and Moonshot AI after Anthropic's 154-page threat intelligence report, published September 10, alleged the two firms routed user requests into Anthropic's Claude models without permission. Anthropic said Moonshot sent more than 23 million exchanges to Claude between May and July and DeepSeek sent over 12.1 million requests in 14 days that July, including queries tied to a municipal Public Security Bureau case-management system and surveillance footage tracking a person across hundreds of police cameras in Chengdu. The regulator initially summoned all seven companies named in the report — including Alibaba, Xiaomi, SenseTime, MiniMax and Zhipu — before narrowing its inquiry to DeepSeek and Moonshot.

China's internet regulator is investigating DeepSeek and Moonshot AI after Anthropic caught both companies quietly funneling millions of user requests, some tied to police and military work, into its Claude models. Here's the part that should stop you: China's own AI champions built their reputations on being the homegrown answer to American AI. Then, according to Anthropic, they spent months secretly asking that same American AI to do the work and serving the answers to their own users as if the work were theirs. Anthropic laid out the case in a 154-page threat intelligence report published on September 10, titled "Detecting and countering misuse of AI." The company says Moonshot AI, maker of the Kimi assistant, routed more than 23 million exchanges to Claude between May and July. DeepSeek, in a tighter window, sent over 12.1 million requests to Claude in just 14 days that July. Anthropic calls the technique illicit distillation: capturing another company's model outputs to train or improve your own, without permission and without telling users what's actually answering them. China's Cyberspace Administration didn't wait long to react. According to The Information and the South China Morning Post, the regulator initially summoned representatives from all seven companies named in Anthropic's report, including Alibaba, Xiaomi, SenseTime, MiniMax and Zhipu, before narrowing its inquiry to DeepSeek and Moonshot specifically. That's the tell. Out of seven firms caught leaning on Claude, only two are now facing a formal data-security probe. The volume numbers are eye-catching, but the content is what worries Beijing. Anthropic's report describes a case in which engineers building a case-management system for a municipal Public Security Bureau had their requests relayed by DeepSeek to Claude, including queries that compared an individual's movements against police records and national identification numbers. In a separate episode, a user Anthropic assessed as likely affiliated with the People's Liberation Army asked Moonshot's Kimi to analyze surveillance footage tracking a person across hundreds of police cameras in Chengdu, some of them positioned outside PLA facilities and defense-linked research institutes. AI Model Distillation Becomes the New Battleground Between the US and China https://startupfortune.com/ai-model-distillation-becomes-the-new-battleground-between-the-us-and-china/ A White House official has accused China's Moonshot AI of building its new Kimi K3 model by distilling Anthropic's Fable, the latest flashpoint in a widening dispute over AI model distillation. OpenAI and Anthropic have separately accused DeepSeek, Moonshot, and MiniMax of extracting capabilities from US models through millions of queries, and... - how to distill AI models cheaply https://startupfortune.com/ai-model-distillation-becomes-the-new-battleground-between-the-us-and-china/ - US China AI model competition regulations https://startupfortune.com/ai-model-distillation-becomes-the-new-battleground-between-the-us-and-china/ That's government surveillance data and military-adjacent footage, potentially sitting on servers Beijing doesn't control. Anthropic also detailed how Moonshot obscured the routing. In one roughly ten-day stretch, the company relayed close to 300,000 customer requests through a proxy network of 5,380 accounts, most of them registered through servers in Singapore and Japan, according to Anthropic's findings as reported by Cyber Kendra and the Free Press Journal. That's not an accidental integration slip. That's infrastructure built to hide where the answers were coming from. For scale, Alibaba's own Claude traffic in the same report ran even higher, over 151 million exchanges between May and July, though Alibaba wasn't swept into the narrowed CAC probe the way DeepSeek and Moonshot were. The timing makes it worse None of this was going to land quietly, but the timing amplifies it. President Trump has said Xi Jinping is arriving for talks on September 24, and the two governments have reportedly been discussing an AI incident-reporting mechanism ahead of that meeting, according to Trivium China. A story about Chinese national-security data leaking to a US frontier lab, breaking just as the two heads of state prepare to sit down over AI policy, is not a coincidence Beijing can shrug off. It's also an awkward moment for the narrative that China's AI sector has closed the gap with American labs on its own terms. DeepSeek spent early 2026 being held up as proof that Chinese labs could match US frontier performance on a fraction of the compute budget. Now the company Anthropic says was quietly asking Claude to do the heavy lifting, on over 12 million occasions in two weeks, is the same one facing a state security review over what got sent along the way. DeepSeek and Moonshot have not issued detailed public responses to Anthropic's specific claims, and neither Alibaba nor the other named firms have said whether they'll face similar scrutiny. What's confirmed is narrower and more concrete: China's regulator has picked two companies out of seven, and it's asking exactly what left their servers, and where it went. Also read: OpenAI Pitches a US-Led AI Rulebook to Counter the UN's Watchdog Plan https://startupfortune.com/openai-pitches-a-us-led-ai-rulebook-to-counter-the-uns-watchdog-plan/ • OpenAI launches GPT-6 Sol and Luna at half the price the same week rivals ship too https://startupfortune.com/openai-launches-gpt-6-sol-and-luna-at-half-the-price-the-same-week-rivals-ship-too/ • Snorkel AI Triples Its Valuation To $3.5 Billion On Surging AI Data Demand https://startupfortune.com/snorkel-ai-triples-its-valuation-to-35-billion-on-surging-ai-data-demand/ Washington threatens to sanction Chinese AI firms over distillation theft as Beijing fires back https://startupfortune.com/washington-threatens-to-sanction-chinese-ai-firms-over-distillation-theft-as-beijing-fires-back/ US Treasury Secretary Scott Bessent threatened sanctions and Entity List blacklisting against Chinese AI firms over industrial-scale distillation campaigns, citing forensic evidence of American model watermarks inside Chinese products. China's commerce ministry called it 'AI hegemonism' Monday and threatened countermeasures, pointing out that US... - how to use Chinese AI models https://startupfortune.com/washington-threatens-to-sanction-chinese-ai-firms-over-distillation-theft-as-beijing-fires-back/ - AI sanctions China trade war https://startupfortune.com/washington-threatens-to-sanction-chinese-ai-firms-over-distillation-theft-as-beijing-fires-back/ This article is posted in AI News https://startupfortune.com/category/ai/ , check it out for more related stories. Join the discussion Open in the community → https://startupfortune.com/community/ Almost there. Sign in and your reply posts straight away.