US artificial intelligence (AI) company Anthropic has accused Chinese e-commerce and technology giant Alibaba of “brazenly” and “illicitly” extracting the capabilities of its Claude AI model, BBC reported.
In a letter sent to two members of the US Congress, the San Francisco-based company alleged that operators linked to Alibaba carried out nearly 29 million interactions with Claude using thousands of fraudulent accounts in what it described as the largest AI model extraction campaign of its kind.
Separately, according to an AP report from Washington, Alibaba has sued the US Department of Defense, seeking removal from the Pentagon’s list of Chinese military companies.
The designation bars listed firms from securing US defence contracts and carries significant reputational consequences.
In a petition filed this week in the San Jose division of the US District Court for the Northern District of California, Alibaba, which is listed on the New York Stock Exchange, argued that the designation, announced on 8 June, “has no basis in fact or law” and that the Pentagon failed to follow a fair process before reaching its conclusion.
The lawsuit is the latest challenge by a Chinese company against the Pentagon over national security-related designations.
Congress directed the Defense Department in 2021 to compile a list of Chinese companies it believes are directly controlled by China’s military or security services, or that contribute to the country’s defence industrial base, amid growing concerns in Washington over China’s military expansion.
Anthropic urged Congress to impose penalties on companies behind such attacks and strengthen measures to prevent US-developed AI technology from being misappropriated.
BBC said it had contacted Alibaba for comment and sought further details from Anthropic.
In the 10 June letter addressed to US Senators Tim Scott and Elizabeth Warren, Anthropic accused Alibaba of conducting “the largest campaign to illicitly extract Claude’s capabilities”.
According to Anthropic, the operation relied on so-called “distillation attacks”, in which outputs from a more advanced AI model are used to train a less capable one.
Anthropic said operators linked to Alibaba specifically targeted Claude’s most valuable features, including its ability to handle longer, more complex tasks and its decision-making processes. It described such attacks as being conducted on an “industrial scale” to harvest and repackage American AI capabilities.
The company also cited other alleged attacks that it said posed potential risks to US military interests.
“Distillation attacks turn hundreds of billions of dollars in American investment and [research and development] into a massive subsidy for our geopolitical competitors,” Anthropic said.
It also referred to Pentagon claims that Alibaba, along with companies including BYD and Baidu, has links to the Chinese military.
Those companies have denied the allegations, while Alibaba is seeking to overturn its designation through the courts.
US AI developers have previously accused Chinese competitors of using distillation techniques to develop rival models at a fraction of the cost of American systems. OpenAI has also made similar allegations against Chinese groups.
Anthropic is one of the leading AI developers and, alongside ChatGPT maker OpenAI, is preparing for a widely anticipated stock market debut that could make it one of the world’s most valuable companies.
However, some of Anthropic’s more advanced AI models, including Mythos, have also drawn cybersecurity scrutiny over their ability to identify vulnerabilities in computer systems.
