China is investigating DeepSeek and Moonshot, The Information reports

While the CAC summoned all 7 companies namechecked by Anthropic’s report, the probe quickly zeroed in on DeepSeek and Moonshot due to the examples the report detailed.

Jing Yang wrote that on Tuesday, about her own story. She and Qianer Liu reported for The Information that the Cyberspace Administration of China is investigating both companies over user data that may have reached Anthropic.

Regulators have questioned staff at both. The probe is open and no penalty has been decided. Neither company has commented publicly, and TNW has not independently verified the report.

The Cyberspace Administration is China’s internet regulator. It licenses AI services for the domestic market, polices what leaves the country in the way of data, and can order products withdrawn. A summons from it is not a court proceeding, and it does not need one to act.

What set it off

The trigger was an American company’s document. TNW covered Anthropic’s threat intelligence report on 10 September.

In it, Anthropic said seven China-based labs had been relaying customer requests through Claude to harvest its output. It calls the practice illicit distillation. The seven are Alibaba, Moonshot AI, DeepSeek, Zhipu, MiniMax, Xiaomi and SenseTime.

The report covers misuse Anthropic says it found between December 2025 and August 2026. It logs each campaign under its own case code and describes how the traffic was detected, including the use of fraudulent accounts to spread requests across many identities.

The volumes in the published report come to roughly 190 million exchanges. They are not evenly spread.

Alibaba accounts for more than 151 million, logged between May and July. Moonshot is next at about 23 million across the same months. DeepSeek’s campaign is smaller in total but far denser: more than 12.1 million exchanges inside 14 days in July. Zhipu runs to 3.4 million, and Xiaomi to just over 400,000.

The Moonshot campaign gives a sense of the method. Anthropic counted almost 300,000 requests in a single ten-day window, spread across 5,380 fraudulent accounts. It said Moonshot forwarded the requests silently and showed users the answers as though its own Kimi model had written them.

All seven were summoned. The investigation has settled on the second and third largest.

For Alibaba this is the second accusation in three months. In June, Anthropic accused Alibaba of distilling Claude into its Qwen models, and TNW reported it at the time. Three weeks after that, Alibaba moved to ban Claude Code internally over what it called a backdoor risk to Chinese users.

Why those two

Jing Yang says the probe narrowed because of the examples Anthropic detailed, rather than the totals.

One of those examples, according to The Information, involved a user Anthropic assessed as likely affiliated with the People’s Liberation Army. That user asked Moonshot’s Kimi to analyse surveillance footage following a person across hundreds of police cameras in Chengdu, including cameras outside PLA facilities and defence-linked institutes. Moonshot allegedly passed the request and the footage to Claude without telling the user.

Police camera footage and material tied to defence institutes sit in the categories China’s own cross-border data rules treat most carefully. A lab accused of moving that abroad has a domestic problem before it has an international one.

Our September coverage recorded a related detail from the same report. One relayed query exposed live credentials tied to a Russian defence agency.

Zhipu, a third name on Anthropic’s list, spent last week on an unrelated data problem of its own. It apologised on Monday after a developer found its ZCode tool uploading encrypted snapshots of local code repositories without consent.

Xiaomi, a fourth, released the highest-scoring open-weight model yet published on Tuesday. It has not responded to the Anthropic allegation either.

The complaint has flipped direction

Anthropic’s grievance was about what left Claude. Chinese labs, it said, were taking its model’s output and passing it off as their own.

Beijing’s grievance is the opposite. The regulator’s concern is what went the other way: Chinese users’ data landing on an American company’s servers without those users knowing.

Same alleged conduct, read from the other end. China is not endorsing Anthropic’s complaint. It has found its own inside the same evidence.

Twelve days ago, Beijing called this unfounded

On 10 September, the day Anthropic published, China rejected a US intelligence advisory on Chinese AI distillation as unfounded. The foreign ministry dismissed the accusations in public.

Twelve days later its internet regulator had summoned every company Anthropic named.

Those are two separate documents. The US advisory, published on 9 September by the NSA, FBI and CISA, named six Chinese AI firms. Anthropic’s report, out the following day, described seven. Beijing’s public position addressed the first.

Bad timing for both

DeepSeek is due to brief the Security Council this week on the risks AI poses to international security. TNW reported that on Tuesday. The briefing falls during the UN General Assembly, and Moonshot is among the other companies appearing.

Moonshot has a different problem. It is working towards a Hong Kong listing, and an open regulatory investigation is the kind of thing that has to appear in a prospectus.

Neither company has said anything about the probe. Anthropic has not commented on the fact that its report has prompted a foreign regulator to summon seven companies.

Chinese regulators have not set a timetable and have not said whether they will penalise either company. The largest campaign in Anthropic’s report belongs to a company that is not, for now, the subject of the inquiry.