Anthropic sent a letter to the U.S. Congress, accusing Chinese tech giant Alibaba of using large-scale "distillation attacks"

TechFlow reports, August 3, according to Forbes, Anthropic recently sent a letter to the U.S. Congress, accusing Chinese tech giant Alibaba of illegally stealing AI intellectual property from its large language model Claude through large-scale "Distillation Attack".

Earlier reports indicated that Alibaba is suspected of registering approximately 25,000 fake accounts, launching 28.8 million prompt-response interactions within three months, and cumulatively extracting approximately 57.6 billion tokens of digital knowledge. The obtained data is suspected to be used to train Alibaba's own large language model, focusing on advanced AI capability areas such as agentic reasoning. It is reported that the above operations were deliberately designed to simulate normal user behavior to evade detection.