Breaking

China Regulator Probes DeepSeek and Moonshot After Anthropic Data-Routing Allegations

September 22, 2026 at 14:00 EDT

Policy & Regulation
Security
Foundation Models

China's internet regulator is investigating DeepSeek and Moonshot AI after Anthropic alleged that the companies routed sensitive user requests to Claude without customers knowing. The inquiry is examining whether data tied to Chinese military, police and state-owned corporate users was sent to Anthropic's systems.

September 2026 · China internet regulator

AI routing allegations become a data-security probe

DeepSeek and Moonshot AI are under scrutiny after allegations that sensitive customer prompts were silently sent to Claude, including data tied to military, police and state-owned corporate work.

companies under probe

300k

Moonshot requests in one 10-day episode

12.1M+

DeepSeek-attributed exchanges over 14 days

Chinese AI firms named by U.S. agencies

Reported distillation scale

Stacked blocks compare exchange counts reported for Moonshot and DeepSeek; the time windows differ.

Alleged routing path

User prompt

Sensitive work entered a domestic AI service.

Service endpoint

The user expected Kimi or DeepSeek.

Proxy accounts

Traffic moved through intermediary accounts.

Claude response

Answers and traces were allegedly captured.

Why users care

Low-cost Chinese models remain attractive for coding and general AI work, but buyers now need clearer proof of where prompts are processed.

What regulators test

The core issue is whether military, police and state-owned corporate data crossed provider boundaries without user knowledge.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.