China Opens Investigation Into DeepSeek and Moonshot Over Alleged Data Leaks to Anthropic's Claude
Key Takeaways
- •Chinese regulators have opened an investigation into DeepSeek and Moonshot AI over allegations that the firms secretly routed user data to Anthropic.
- •Officials visited both companies to question executives and staff about whether police, military, and state-linked corporate data ended up on U.S. servers.
- •The probe was prompted by Anthropic's September 10 threat intelligence report, which accused seven China-based AI labs of distilling Claude's outputs through fraudulent accounts.
- •Anthropic's report claimed Moonshot routed more than 23 million exchanges to Claude through 5,380 fraudulent accounts, while DeepSeek generated over 12.1 million exchanges in a 14-day window in July.
- •Moonshot's confidentially filed Hong Kong IPO, targeting $3 billion at a $50 billion valuation, would need to disclose the investigation in its prospectus before the listing can proceed.
China is turning Anthropic's own leak-detection playbook against two of its most prominent AI labs. The Cyberspace Administration of China (CAC), the country's top internet regulator, has opened an investigation into DeepSeek and Moonshot AI over claims that both companies secretly routed sensitive user data to Anthropic, the company behind Claude, The Information reported Tuesday, citing people familiar with the matter.
DeepSeek, the Hangzhou-based lab whose low-cost models rattled global markets in early 2025, and Moonshot, the Beijing-based company behind the Kimi chatbot, rank among China's best-known AI developers.
Officials reportedly visited the offices of both companies to interview executives and staff. Regulators want to know whether sensitive police, military, and state-linked corporate data ended up on U.S. servers.
For most of 2026, Anthropic has accused Chinese labs of stealing Claude's outputs to train cheaper rivals. The company has even claimed these startups were rerouting their queries to Anthropic, effectively giving users Claude-grade answers at a fraction of the price. Beijing is now investigating whether those claims are true.
The trigger was Anthropic's fourth threat intelligence report, a 154-page document published September 10. It accused seven China-based labs—Alibaba, Moonshot, DeepSeek, Zhipu (Z.ai), MiniMax, SenseTime, and Xiaomi—of what Anthropic calls "illicit distillation."
Distillation, in AI shorthand, refers to training a smaller model on a larger model's outputs. Anthropic maintains that the practice is legitimate on its own; what it objects to is carrying it out through fraudulent accounts. According to the report, Moonshot routed more than 23 million exchanges to Claude through 5,380 fraudulent accounts, while DeepSeek generated more than 12.1 million exchanges in a 14-day window in July.
The CAC initially summoned representatives from all seven companies named in the report. It has since narrowed its focus to DeepSeek and Moonshot specifically, and no penalties have been announced.
The timing is awkward for both firms. DeepSeek is scheduled to brief the United Nations Security Council on AI risks this week, sharing a stage with Anthropic CEO Dario Amodei. The session lands just one day before President Donald Trump's September 24 summit with Xi Jinping, where AI is reportedly on the agenda.
Moonshot faces complications of its own. The company confidentially filed for a Hong Kong IPO on September 3, targeting $3 billion at a $50 billion valuation—and any disclosed regulatory investigation would need to appear in that prospectus before the listing can move forward, tying the offering's path directly to how the investigation unfolds.
