DeepSeek and Moonshot AI under investigation by China over data security

AI data security model distillation DeepSeek Moonshot AI Anthropic

The Cyberspace Administration of China (CAC) has launched an investigation into AI companies DeepSeek and Moonshot AI following allegations from the U.S.-based firm Anthropic. Anthropic’s September 2026 threat intelligence report accused seven Chinese AI laboratories—including Alibaba, Zhipu, MiniMax, SenseTime, and Xiaomi—of engaging in illicit distillation. This process involves using the outputs of a more advanced model, such as Anthropic’s Claude, to train smaller, less expensive models.

Anthropic reported that Moonshot AI allegedly routed approximately 23 million user prompts to Claude between May and July 2026, often using thousands of fraudulent accounts. DeepSeek is accused of rerouting over 12.1 million queries to Claude during a 14-day period in July. The report claims these interactions included sensitive information, such as data from a Chinese municipal police system and a Russian government agency.

While Anthropic’s focus is on the theft of its model's intellectual property and reasoning capabilities, Chinese regulators are investigating the potential for data leaks. The CAC is examining whether sensitive Chinese military, police, and corporate data was inadvertently transferred to U.S. servers through these interactions, which would violate China’s data security laws. Officials have reportedly interviewed executives and employees at both DeepSeek and Moonshot AI as part of the probe.

Entities

Anthropic · Claude · Cyberspace Administration of China · DeepSeek · Moonshot AI

Claims

What the coverage asserts, and how many sources carry each claim.

[● 7 SOURCES] The investigation aims to determine if sensitive Chinese military, police, or corporate data was transferred to U.S. servers. sputniknews.cn · donanimgunlugu.com · decrypt.co · thenextweb.com · phocapblockchain.net · +2 more
[● 12 SOURCES] The Cyberspace Administration of China (CAC) has launched an investigation into DeepSeek and Moonshot AI regarding data security. donanimgunlugu.com · sputniknews.cn · borncity.com · www.cryptopolitan.com · decrypt.co · +7 more
[● 8 SOURCES] Chinese regulators have visited the offices of DeepSeek and Moonshot AI to interview executives and employees. sputniknews.cn · donanimgunlugu.com · decrypt.co · thenextweb.com · phocapblockchain.net · +3 more
[● 11 SOURCES] Anthropic accused seven Chinese AI companies of using illicit distillation to train their models using Claude's outputs. dijitaliyidir.com · borncity.com · www.cryptopolitan.com · decrypt.co · phocapblockchain.net · +6 more
[● 8 SOURCES] Moonshot AI allegedly forwarded approximately 300,000 customer requests to Claude using thousands of fraudulent accounts. dijitaliyidir.com · www.cryptopolitan.com · decrypt.co · donanimgunlugu.com · www.elperiodico.digital · +3 more
[● 6 SOURCES] DeepSeek reportedly rerouted over 12.1 million user queries to Claude within a 14-day period in July. dijitaliyidir.com · www.cryptopolitan.com · decrypt.co · www.elperiodico.digital · www.drweb.de · +1 more