Anthropic says DeepSeek, other Chinese AI firms scraped Claude to train rival models

undefined 24, 2026

Source: interestingengineering

Anthropic claims Chinese AI firms DeepSeek, MiniMax, and Moonshot used fake accounts and proxy networks to extract over 16 million interactions from its Claude chatbot to train rival models. Anthropic says the large-scale “distillation” violated its terms and aimed to replicate advanced capabilities. The allegations follow similar warnings from OpenAI, heightening US concern over Chinese firms leveraging American AI. Anthropic says it has tightened safeguards and is sharing threat intelligence across the industry.

Share on