NEW DELHI — The U.S. administration plans to expand cooperation with domestic artificial intelligence companies to counter what officials describe as “industrial-scale campaigns” by foreign entities, primarily based in China, to extract and replicate advanced technologies, according to a report.
Michael Kratsios, director of the White House Office of Science and Technology Policy, said in an internal memo that new intelligence indicates foreign actors are exploiting U.S. AI systems through a method known as “distillation,” the BBC reported.
Kratsios said the strategy is designed to undermine American research and development by gaining access to proprietary information and applying it to competing systems.
To address the issue, the White House plans to share more intelligence with U.S. AI companies on the tactics and actors involved in such efforts, while improving coordination to detect and prevent breaches. Officials are also working on a set of best practices to help companies identify, mitigate, and respond to these activities, and are exploring ways to hold those responsible accountable.
Distillation campaigns typically involve operating thousands of user accounts on AI platforms to mimic normal usage patterns. These accounts then attempt to “jailbreak” systems or extract restricted information, which can be used to train rival AI models.
“As methods to detect and mitigate industrial-scale distillation grow more sophisticated, foreign entities who build their AI capabilities on such fragile foundations should have little confidence in the integrity and reliability of the models they produce,” Kratsios said.
China rejected the allegations. A spokesperson for its embassy in Washington said the country’s technological progress is driven by “its own dedication and effort as well as international cooperation.”
The issue has drawn attention from the private sector as well. In March, U.S.-based AI company Anthropic accused three Chinese firms—DeepSeek, Minimax, and Moonshot AI—of illegally extracting capabilities from its Claude model to enhance their own systems. (Source: IANS)
