Bessent says US needs more open-source AI models to compete with China

The excerpt does not state the date of the remarks or the specific setting within Congress.

According to the excerpt, Bessent said Chinese labs have trained open-source models on U.S. technology companies' closed-source models through a process known as distillation. He described distillation as 'a polite, scientific word for steal.' The excerpt does not name the Chinese labs involved, does not identify which U.S.

The excerpt does not include comment from Chinese labs, from U.S. technology companies, or from other officials responding to the remarks. It also does not indicate whether any legislation on open-source models is under consideration.