By Blessing Nweke
US artificial intelligence company Anthropic has accused Chinese technology giant Alibaba of carrying out what it described as the largest known effort to unlawfully extract the capabilities of its Claude AI model, escalating tensions over the protection of advanced AI technology.
In a letter dated June 10 and sent to US Senators Tim Scott and Elizabeth Warren, the San Francisco-based AI developer alleged that operators linked to Alibaba conducted nearly 29 million interactions with Claude through thousands of fraudulent accounts.
Anthropic claimed the activity constituted a large-scale “distillation attack,” a technique in which outputs from a more advanced AI model are used to train a less capable rival system. According to the company, the campaign focused on extracting Claude’s most valuable features, including its ability to handle lengthy and complex tasks and its decision-making processes.
The company warned that such attacks are being conducted on an “industrial scale,” enabling foreign competitors to replicate expensive American AI research and development at a fraction of the cost. Anthropic urged Congress to strengthen protections for US technology and impose penalties on organizations involved in such activities.
The allegations add to growing concerns in Washington over the transfer of advanced AI capabilities to geopolitical rivals. Anthropic argued that distillation attacks effectively transform billions of dollars of US investment into a subsidy for competing nations.
The letter also referenced Pentagon claims linking Alibaba and other Chinese firms, including BYD and Baidu, to China’s military. Those companies have denied such allegations. Alibaba recently filed a lawsuit seeking removal from a US Defense Department blacklist.
Alibaba has not publicly responded to Anthropic’s latest accusations. Anthropic’s claims follow similar concerns previously raised by OpenAI regarding alleged AI model distillation by Chinese groups.
Original article
Comments