US artificial intelligence (AI) giant Anthropic has accused Chinese e-commerce and technology company Alibaba of “brazenly” and “illegally” extracting features from Claude’s AI model.
The San Francisco-based company said in a letter to two U.S. members of Congress that operators associated with Alibaba conducted approximately 29 million interactions with Claude using thousands of fraudulent accounts in what it called the largest exploitation campaign of its kind.
Mr. Anthropic called on Congress to punish the companies behind such attacks and strengthen measures to prevent American technology from being stolen.
The BBC contacted Alibaba for comment and asked Anthropic for further details.
Mr. Anthropic’s letter, dated June 10, to Tim Scott and Sen. Elizabeth Warren, accused Alibaba, which is listed on the New York Stock Exchange, of waging “the largest campaign ever to unfairly exploit Mr. Claude’s abilities.”
According to Anthropic, the campaign was run through a so-called “distillation attack,” which extracts answers from a more powerful AI model to train a weaker AI model.
Anthropic said Alibaba-related carriers targeted Claude’s most valuable skills, including his ability to take on longer and more complex tasks and his approach to decision-making.
These types of attacks are carried out on an “industrial scale” to allow Chinese companies to harvest and repackage U.S. AI capabilities as their own, the company said.
The letter also cited other alleged attacks, saying Anthropic posed a threat to the U.S. military.
“The distillation attack generated hundreds of billions of dollars in American investment, [research and development] “It’s a huge subsidy to geopolitical competitors,” Anthropic said.
The newspaper cited claims by the US Department of Defense that Alibaba and several other major companies, including automaker BYD and tech company Baidu, have ties to the Chinese military.
Both companies deny the allegations, but Alibaba sued the U.S. government this week to remove its name from the Pentagon’s blacklist.
U.S. developers have previously accused Chinese competitors of using distillation attacks to train models comparable to U.S. AI techniques at a fraction of the cost.
OpenAI has previously accused Chinese groups of employing similar practices.
Anthropic is a leading AI developer and is gearing up for a major stock market debut that could make it one of the world’s most valuable companies, along with ChatGPT creator OpenAI.
But some of Anthropic’s more advanced models, such as Mythos, have raised cybersecurity concerns about their ability to target weaknesses in computer systems.
