China investigates DeepSeek and Moonshot over alleged Claude data routing

China’s internet regulator is investigating DeepSeek and Moonshot AI after Anthropic alleged that the two Chinese AI developers secretly sent some customer requests to its Claude model. The allegation began as a fight over “distillation”—using a stronger model’s answers to help train a rival—but it raised a more serious question: whether users’ prompts, including potentially sensitive information, were sent to a foreign service without their knowledge. Anthropic said the firms used large networks of fraudulent accounts to access Claude and that some traffic may have included sensitive Chinese corporate and state-linked data. The inquiry turns what might have been an argument over copying AI capabilities into a test of where Chinese data goes and what AI providers disclose about the systems operating behind their apps.