Anthropic Accuses Three Chinese AI Companies of Industrial-Scale Model Distillation Campaigns

Anthropic has escalated the tensions in the global AI race by formally accusing three Chinese AI companies of running coordinated campaigns to extract information from its models, marking the latest American tech firm to level such claims after OpenAI issued similar complaints. The allegations center on what Anthropic describes as industrial-scale distillation efforts, where Chinese companies allegedly used sophisticated techniques to probe and extract knowledge from Claude and other frontier models. This follows closely on the heels of Reuters reporting that DeepSeek trained its latest AI model on Nvidia's best chips despite US export bans, with the chips reportedly located in Inner Mongolia. The US government believes DeepSeek will remove technical indicators that might reveal its use of American AI chips. The situation has created a significant diplomatic headache, with implications for US-China technology relations that extend far beyond the AI industry.