Anthropic Says DeepSeek, MiniMax Distilled AI Models for Gains

Anthropic accused three top Chinese AI developers—DeepSeek, MiniMax, and an unnamed third—of illicitly extracting outputs from its models to enhance competing products, escalating U.S. concerns over IP theft in AI. The allegations, revealed in February 2026, claim systematic "distillation" of Anthropic’s proprietary AI responses to train or fine-tune rival systems, bypassing costly R&D. This follows broader U.S. warnings about Chinese firms leveraging foreign AI advancements through unauthorized means, intensifying geopolitical tensions in tech competition. Anthropic’s statement stops short of legal action but signals growing industry pushback against cross-border model exploitation, particularly in high-stakes generative AI. The incident underscores rising risks of AI model reverse-engineering as firms race for dominance, with distillation emerging as a contested shortcut in global AI development.