China's internet regulator probes DeepSeek and Moonshot over alleged data routing to Claude
China's Cyberspace Administration is probing DeepSeek and Moonshot AI over Anthropic's allegation that both routed millions of user exchanges through Claude to train their own models.
Why it matters
The questions center on whether Chinese state-linked and police data reached U.S. servers, and the timing lands on two busy moments: Moonshot has filed confidentially for a $3 billion Hong Kong IPO and DeepSeek is briefing the UN Security Council on AI risks this week. One outlet carries the story, no regulator has confirmed it on the record in the available evidence, and no company response is reported.
Key facts
China's Cyberspace Administration opened an investigation into DeepSeek and Moonshot AI.
Reported by Decrypt
Officials visited the offices of both companies to interview executives and staff.
Reported by Decrypt
Regulators want to know whether police, military and state-linked corporate data ended up on U.S. servers.
Reported by Decrypt
Anthropic's September 10 report alleged that DeepSeek and Moonshot AI secretly routed user exchanges through Claude to train their own models.
Reported by Decrypt
Anthropic alleges Moonshot AI routed more than 23 million exchanges to Claude through 5,380 fraudulent accounts.
5,380 fraudulent accountsReported by Decrypt
Anthropic alleges DeepSeek generated more than 12.1 million exchanges in a 14-day window in July.
Reported by Decrypt
Moonshot AI filed confidentially for a $3 billion Hong Kong IPO.
3,000,000,000 USDReported by Decrypt
What happened
China's Cyberspace Administration has opened an investigation into DeepSeek and Moonshot AI, Decrypt reported, citing The Information and people familiar with the matter. Decrypt said officials visited the offices of both companies to interview executives and staff, and that regulators want to know whether sensitive police, military and state-linked corporate data ended up on U.S. servers. The probe follows Anthropic's September 10 report, which alleged the two labs secretly routed user exchanges through Claude to train their own models; Decrypt reported that regulators first summoned all seven companies named in that report before narrowing the investigation to DeepSeek and Moonshot. Anthropic's figures, as relayed by Decrypt, put Moonshot at more than 23 million exchanges routed through 5,380 fraudulent accounts, and DeepSeek at more than 12.1 million exchanges in a 14-day window in July. The available reporting records no response from DeepSeek or Moonshot, and does not say whether either company disputed the numbers Anthropic published.
How the story developed
- Update
Anthropic publishes report alleging AI labs routed user exchanges to Claude
The report named seven companies and included allegations about DeepSeek and Moonshot AI.
- First report
First report by Decrypt
China Probes DeepSeek and Moonshot Over Alleged Data Leaks to Anthropic's Claude
- Update
The Information reports the Cyberspace Administration opened an investigation
Decrypt cited The Information, which cited people familiar with the matter, and reported that officials interviewed executives and staff at both companies.
- Update
Probe narrowed from seven companies to DeepSeek and Moonshot AI
Regulators summoned all seven companies named in Anthropic's report before focusing on the two labs.
- Update
Moonshot AI confidentially files for a $3 billion Hong Kong IPO
The filing was reported as confidential.
- Update
DeepSeek briefs the UN Security Council on AI risks
Decrypt said the briefing is scheduled for the week of its report.
Affected entities
Original sources
| Publisher | Report | Role | Published |
|---|---|---|---|
| Decrypt Crypto media | China Probes DeepSeek and Moonshot Over Alleged Data Leaks to Anthropic's Claude | Original |