By Eduardo Baptista
BEIJING, Oct 16 (Reuters) – China proposed export controls on 23 AI technologies on Friday, including those involving models, agents and computing clusters, its latest move to limit transfers of tools it views as vital to national security and geopolitical competition with Washington.
The proposals cover overseas deployment of AI models containing restricted technology, but exclude access through application programming interfaces (APIs), websites, apps and software-as-a-service platforms, China’s commerce ministry said.
The distinction could allow Chinese AI developers such as DeepSeek and Alibaba to continue expanding internationally while giving Beijing greater oversight of transfers of sensitive AI technology abroad.
Chinese AI models have gained ground abroad on the back of lower prices and improving capabilities since DeepSeek’s R1 model emerged early last year, raising the stakes for Beijing as it decides how tightly to guard the sector.
Reuters reported in July that Chinese authorities had discussed with major technology companies, including Alibaba, ByteDance and Z.AI, potentially restricting overseas access to China’s most advanced AI models, including unreleased ones.
PREVENTING AI FROM HARMING NATIONAL SECURITY
The capabilities of frontier models like Anthropic’s Mythos have highlighted the potential for AI to turbocharge cyberattacks, worrying Beijing and prompting Chinese companies to seek to develop their own versions.
Washington’s export controls on China’s access to cutting-edge American AI chips needed to train frontier models have so far helped keep US firms Anthropic and OpenAI in the lead, though Chinese rivals are finding workarounds to narrow the gap.
While both countries strive to lead in AI, some bilateral efforts are underway to address existential risks posed by the technology. The two powers have agreed to keep humans in control of nuclear-weapons decisions and last month set up a dialogue and an incident-communications channel for advanced AI.
The commerce ministry said on Friday the controls were partly intended to prevent misuse of AI technologies, including systems capable of autonomously generating and modifying malicious software that could threaten critical infrastructure and national security.
It said most AI technologies would remain freely exportable, with simplified licensing for open-source technology exports and Chinese companies operating their own computing centres overseas. Technology already in the public domain through open-source releases or academic papers would be outside the proposed controls.
BROADER REVISION
The proposals form part of a broader revision of China’s technology export catalogue, covering 40 technologies across AI, biotechnology and solar power.
The proposals also include prohibiting the transfer of intellectual property ownership involving some novel cell and gene therapies and small interfering RNA (siRNA) technologies.
Exporters must register certain overseas technology transfers from Friday, pending final rules after public consultation on the proposals closes November 15.
(Reporting by Eduardo Baptista; Additional reporting by Shi Bu, Liz Lee and Andrew Silver; Editing by Alison Williams and Philippa Fletcher)
Comments