The United States may increase pressure on Chinese artificial intelligence companies as it considers applying sanctions against AI models suspected of using intellectual property developed by American companies. The signal was given by Treasury Secretary Scott Bessent amid the advance of open-source models created in China, which have been gaining ground and increasing competition with the leading U.S. companies.
During an interview with Fox Business, Bessent said that the American government intends to investigate whether Chinese models were trained with technology obtained improperly.
“We have seen a lot of talk about open-source models emerging and threatening large language models in the U.S.,” Bessent said. “This administration supports open-source models, but what we do not support is the theft of intellectual property. If we see, especially, that foreign models are stealing from our large companies, we have the ability to sanction them for that theft.”
The statement comes at a time of rapid evolution in Chinese artificial intelligence. One of the highlights is Kimi K3, developed by Moonshot AI, which gained notoriety for its performance and began competing directly with models from companies such as OpenAI and Anthropic.
On Monday, information also emerged that the administration of current U.S. president Donald Trump is considering stricter measures against Chinese open-source models. Among the possibilities under discussion is a broad restriction on access to these technologies in the American market, although the issue still generates disagreements.
The possible adoption of sanctions represents another step in the United States' strategy to maintain leadership in the race for artificial intelligence. In recent years, Washington has already restricted China's access to advanced chips, strengthened export controls, and intensified actions to protect technologies considered strategic.
Another point that divides opinions is the so-called model distillation, a technique that allows part of the capabilities of a larger AI system to be transferred to a smaller and more efficient model. While some companies consider this practice a violation of intellectual property, experts argue that it does not always constitute technology theft.
Microsoft CEO Satya Nadella criticized this interpretation by stating: "Although the great innovation that comes from model providers having fair use rights to train models on public data is necessary, I think it is ironic that the status quo is then to impose restrictive terms on distillation."
At the same time, disputes involving copyright continue to grow. This week, Anthropic received authorization to begin payments to authors after a US$ 1,5 bilhão agreement, resulting from a court decision related to the storage of millions of copyright-protected books for training its artificial intelligence systems.
In addition to the discussions about distillation, part of the sector assesses that the advance of Chinese AI is also the result of investment, its own research, and the strengthening of the open-source ecosystem, factors that have been reducing the technological gap between Chinese and American companies.