Anthropic accuses Alibaba of AI model copying

U.S.-based AI company Anthropic has accused Chinese technology and e-commerce giant Alibaba of attempting to illegally copy the capabilities of its Claude artificial intelligence model.

According to a June 10 letter obtained by The Wall Street Journal, Anthropic told U.S. Senators Tim Scott and Elizabeth Warren that Alibaba and its AI division carried out what it described as the "largest known distillation campaign" against the company to date, CE Report quotes Anadolu Agency.

Anthropic stated that Claude is not available to organizations in China and alleged that Alibaba created approximately 25,000 fake accounts to gain access to the model, generating around 29 million interactions with it.

The company claimed that the activity specifically targeted Claude's most advanced capabilities, including autonomous task planning and execution, software engineering, and long-term planning functions.

In the letter, Anthropic called for tighter export controls on advanced AI chips and new regulations that would impose sanctions on companies found to be engaging in unauthorized model-copying activities in order to protect U.S. artificial intelligence technologies.

Alibaba has not yet responded publicly to the allegations.

On April 23, the U.S. administration said it had evidence that foreign actors, particularly China-based entities, were conducting "industrial-scale" distillation campaigns aimed at replicating American AI technologies.

In artificial intelligence, "distillation" refers to a technique in which a smaller or less capable model is trained using the outputs generated by a more advanced model.

Photo: Chat GPT