A top White House tech official has accused China's Moonshot AI of engaging in a "large-scale, covert industrial distillation" to steal proprietary technology from U.S. artificial intelligence labs. Michael Kratsios stated Wednesday that Moonshot AI distilled output from Anthropic's sophisticated AI large language model, Fable, to aid in the development of its latest Kimi K3 release.
"Undermining American research is unacceptable," Kratsios said in a post on X. He added that the action was a deliberate effort to steal U.S. technology.
Moonshot AI, a Chinese artificial intelligence startup, unveiled its Kimi K3 model on Friday. The company described it as the world's largest open-weight AI system, with 2.8 trillion parameters, and stated that its performance approaches that of Anthropic's Fable model. Distillation is a process where smaller AI models are trained using the output of larger, more expensive ones, serving as a method to reduce the cost of developing powerful new AI tools.
The launch of Kimi K3 comes roughly a month after the U.S. government abruptly withdrew Anthropic's Fable and Mythos models due to security concerns. This development highlights the rapid pace at which China's open AI ecosystem is closing the gap with leading U.S. systems.
In April, the U.S. State Department initiated a global campaign to draw attention to what it described as widespread attempts by Chinese companies, including Moonshot AI, to pilfer intellectual property from American AI laboratories. This information was revealed in a diplomatic cable seen by Reuters.