The US artificial intelligence company Anthropic has accused Chinese tech giant Alibaba of orchestrating a brazen, systematic theft of its AI model Claude – a campaign it says involved almost 29 million fraudulent exchanges. In a letter sent to US senators Tim Scott and Elizabeth Warren on 10 June, the San Francisco-based firm alleged that operators linked to Alibaba used thousands of fake accounts to extract Claude’s most valuable capabilities, including its ability to handle long, complex tasks and its decision-making processes. Anthropic called it “the largest campaign to illicitly extract Claude’s capabilities” and urged Congress to penalise the companies behind such attacks and to strengthen protections against US technology being stolen.
The technique, known as a “distillation attack”, involves extracting answers from a more advanced AI model to train a weaker one. Anthropic said that the campaign was carried out on an “industrial scale” to enable Chinese firms to harvest and repackage American AI as their own. “Distillation attacks turn hundreds of billions of dollars in American investment and [research and development] into a massive subsidy for our geopolitical competitors,” the company wrote. The letter also cited other alleged attacks that it said posed a threat to the US military, referencing the US Department of Defense’s claims that Alibaba – along with car maker BYD and tech company Baidu – are tied to the Chinese military. The companies have denied those allegations, and Alibaba this week sued the US government in an attempt to have its name removed from the Pentagon blacklist.
“US AI firm Anthropic accused Alibaba of orchestrating 29 million illicit data exchanges to steal its Claude model's capabilities.”
The accusation is the latest in a series of claims by leading US AI developers that Chinese rivals are using distillation to close the gap at a fraction of the cost. OpenAI has previously accused Chinese groups of the same practice. Anthropic, a leading AI developer alongside ChatGPT-maker OpenAI, is preparing for a blockbuster stock market debut that could make it one of the most valuable companies in the world. However, some of its more advanced models, such as Mythos, have raised cybersecurity concerns over their ability to target weaknesses in computer systems. The BBC has contacted Alibaba for comment and requested more details from Anthropic.