@kimmonismus: China drafts export rules for its biggest AI models

China's commerce and science ministries published a draft on 16 October that would require a licence to export the weights, code or training methods of the very largest models: 2 trillion or more parameters, very large training runs, or a global top-10 ranking. Weights already published as open source are exempt, new open-weight releases get an automatic licence after registration, and using these models through an API or app isn't affected. Comments close on 15 November.

Why it mattersIf you build on Chinese open models such as DeepSeek, Qwen or Kimi, nothing changes today. It's worth watching if you plan to self-host the biggest future releases.

Curated by Sara Simeone as part of our human editorial coverage. The original sources are linked below; curation is distinct from a Hype or Fact? assessment.

How we knowCompany figures are labelled as their own claims; anything we couldn't verify is labelled as a report or a claim.