China's internet watchdog is investigating DeepSeek and Moonshot AI for potential data leaks to Anthropic, according to The Information, after Anthropic accused the two Chinese AI firms of silently routing user requests to Claude.
The Information reported Tuesday that the Cyberspace Administration of China (CAC), widely considered one of the world's most powerful digital censorship agencies, has summoned representatives from seven Chinese AI companies named in an Anthropic report - DeepSeek, Moonshot, Alibaba, Z.ai, SenseTime, MiniMax, and Xiaomi.
The investigation soon zeroed in on DeepSeek and Moonshot, according to the report. The probe remains ongoing, and regulators have yet to decide whether to penalize the companies.
The scrutiny follows Anthropic's allegations that DeepSeek and Moonshot secretly relayed user requests to Claude and displayed the responses to users as their own.
Some of those requests fed to Claude contained sensitive information, according to a 154-page report Anthropic published earlier this month. One user, who Anthropic said was likely affiliated with China's People's Liberation Army, used what they believed was Moonshot's Kimi model to load surveillance data from a CCTV archive covering hundreds of cameras in Chengdu, the capital of southwestern China's Sichuan province.
DeepSeek also allegedly relayed requests from a user working with Russian Ministry of Defense data, exposing live credentials for a Russian government database, per the report.
"We do not know if Moonshot notified their customers that their requests were being rerouted to Anthropic and exposed to a third party," Anthropic said in its report earlier this month.
Distillation campaigns
Anthropic's allegations came after several U.S. federal agencies alleged that Chinese AI companies systematically extracted capabilities from American AI models through distillation.
China has pushed back against the allegations, calling them groundless and arguing that the U.S. move "politicizes and instrumentalizes distillation," which it said was a "common practice" in the AI field.
The CAC's investigation, however, points to a potentially more sensitive concern for Chinese authorities: whether Chinese AI firms have exposed sensitive military or corporate data to the U.S.
The Latent has reached out to the CAC for comment.