Chinese AI models at the centre of growing divide in Silicon Valley

The United States and China are embroiled in a new technology rivalry centered around artificial intelligence (AI), according to US Treasury Secretary Scott Bessent. Washington is considering restrictions targeting Chinese AI companies and their models, which have become increasingly prominent in the tech industry. US officials have expressed concerns that some Chinese AI models may have been developed using technology copied from American firms.

The US accuses China of illicitly copying proprietary American systems through a process called distillation, which involves one AI model learning from another. Beijing has rejected these accusations, stating that the US allegations lack factual basis and legal grounds. China will take all necessary measures to safeguard its legitimate and lawful rights and interests.

The dispute is dividing the US technology industry, as engineers move seamlessly between US models and Chinese open-weight models at software development platform Replit. Replit CEO and co-founder Amjad Masad emphasized that technology should not be viewed simply in terms of national origin, but rather as an ecosystem. Some groups representing smaller technology companies, such as the Little Tech Association, have urged the Trump administration not to block access to Chinese open-weight AI models, warning that it could drive up prices and reduce innovation.

US officials have accused a Chinese company, Moonshot AI, of improperly distilling an advanced AI model from Anthropic to create its own model. Protecting American companies could be a response to intellectual property theft, but determining when AI model distillation crosses the line into intellectual property theft remains challenging.

Some argue that restricting access to Chinese AI could harm American companies, while others believe it could drive innovation and innovators elsewhere.