Anthropic: Alibaba Stole Claude AI Tech via 29M Exchanges

Artificial intelligence developer Anthropic accuses Chinese technology conglomerate Alibaba of illicitly extracting capabilities from its Claude AI model. Anthropic states that operators linked to Alibaba used about 25,000 fraudulent accounts. These accounts conducted nearly 29 million exchanges with Claude between April 22 and June 5. This activity is described as an industrial-scale distillation attack. This method harvests data from an advanced model to train a weaker system at a fraction of the original cost. Anthropic warns that these attacks turn American investment and research into a subsidy for geopolitical competitors. The company urges congressional intervention to protect domestic AI infrastructure. Alibaba has not publicly responded to these allegations. This situation highlights the ongoing challenges in protecting intellectual property in the rapidly evolving AI landscape.