New Regulations Incoming for China's Advanced Artificial Intelligence Model Exports

New draft rules proposed by China's Ministry of Commerce mandate government approval for the export of advanced artificial intelligence models outside the country. This move shows that a balance between national security risks and global competition is being considered, while casting uncertainty on the country's open-access strategy in artificial intelligence technologies.

The draft rules announced last week by China's Ministry of Commerce stipulate tying the export of the country's most advanced artificial intelligence models to government approval. The Kimi K3 model, launched in July by Chinese startup Moonshot AI and competing with the top systems of American companies, could be downloaded for free worldwide. Chinese companies had gained a strong advantage in shaping the future of this technology by keeping advanced artificial intelligence models easily accessible.

The proposed new regulations stand out as Beijing's move to tighten its control over the global dissemination of the country's high-level artificial intelligence. If the draft rules become law, Chinese companies will have to obtain export licenses for the core technology behind their leading models. Furthermore, while the export of artificial intelligence models that could be used to launch cyberattacks will be banned, models trained with "core data" regarding national security or state secrets will also not be allowed to be taken abroad.

At the same time, government officials tried to reassure users by stating that this step will not block access to Chinese artificial intelligence and that the vast majority of the technology will continue to be freely exported.