Breaking

China Probes DeepSeek and Moonshot AI Over Alleged Claude Data Routing

September 23, 2026 at 14:17 EDT

Foundation Models
Security
Policy & Regulation

China's internet regulator is investigating DeepSeek and Moonshot AI after Anthropic alleged that the companies silently routed customer requests through Claude, raising questions about whether sensitive Chinese government, police and corporate data crossed borders without users' knowledge. The inquiry turns an AI model-distillation dispute into a data-security matter for two of China's most closely watched AI developers.

Sept. 2026 · China AI probe

Claude routing becomes a data-security case

Regulators are examining whether DeepSeek and Moonshot AI sent sensitive customer data to Anthropic's Claude without users knowing, after Anthropic alleged large-scale routing and distillation campaigns.

named companies were summoned before the probe narrowed.

23M+

Moonshot-linked Claude exchanges were observed from May to July.

12.1M+

DeepSeek-linked exchanges were observed in a 14-day July window.

Comparable Claude traffic

The two companies at the center of the probe are linked to different volumes and time windows.

23M+Moonshot

12.1M+DeepSeek

Alleged request path

User request

Prompts, code, credentials or surveillance data enter a local AI service.

→

Silent routing

Selected traffic is allegedly sent through Claude without user notice.

→

Returned answer

Users receive an answer presented as the service they chose.

Why users notice

The episode highlights whether model providers and routers clearly disclose where prompts are processed.

Why regulators notice

Sensitive military, police and state-linked corporate data may trigger cross-border data rules.

Continue reading

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