DeepSeek and MiniMax have become the first major Chinese AI labs to win Beijing’s approval for global use of their models, clearing a path through China’s new AI export rules. The approval was announced late Thursday in a post by Yuyuan Tantian, a social media account affiliated with state broadcaster CCTV, which said both companies had “successfully completed the registration process, as well as the procedures for open-sourcing and authorising overseas use.”
“AI companies can continue their commercialisation efforts as usual,” the post said. In a follow-up post on Friday, the account added that overseas internet users had urged other Chinese open-source models to follow suit.
Neither DeepSeek nor MiniMax immediately replied to Reuters’ request for comment. The approval lands a week after China’s commerce ministry proposed export controls on 23 AI technologies — including model weights and training methods — that had raised fears the country’s open-weight labs could be cut off from the world.
Registration, Not A Ban
The Yuyuan Tantian post framed Beijing’s approach as defining boundaries rather than imposing a blockade. Under the interim rules, transfers of technology ownership, licenses for overseas use, and overseas deployment of unpublished models require registration — but providing API access, web applications or software-as-a-service features is not subject to restrictions, even when powered by restricted underlying models, according to the post.
Technology already in the public domain through open source or academic papers is also exempt. That carveout is what matters most for DeepSeek and MiniMax, whose open-weight releases have driven the surge in Chinese model adoption outside China.
The post drew a pointed contrast with Washington, accusing the US of “using targeted export bans and technology blockades to protect its competitive dominance”. Days earlier, the US had ordered its civilian agencies to remove DeepSeek products from government systems, a move Beijing formally protested.
What It Means For The Open-Weight Labs
DeepSeek is in the middle of a record stretch: it is wrapping up a roughly $15 billion funding round backed by CATL and Tencent, one of China’s largest AI fundraises ever, while simultaneously fending off Washington’s restrictions. Completing the export registration gives the lab a formal green light to keep shipping its models abroad.
MiniMax, which went public in Hong Kong in January, is the other named beneficiary. The approval signals that Beijing’s export regime is a paperwork gateway, not a wall — labs that register and comply can continue operating globally, which preserves the open-weight distribution model that has made Chinese models so widely adopted.
Why It Matters
DeepSeek and MiniMax being first through the gate tells the industry what the new rules actually are. Beijing has chosen to regulate its AI exports through registration and licensing rather than the kind of broad technology bans Washington has pursued. For the thousands of developers and companies building on open Chinese models, that is the difference between a shock and business as usual.
The contrast also sharpens the geopolitical split: the same week the US barred DeepSeek from its federal agencies, China handed it a licence to keep selling everywhere else. Expect the approval to become a template — and for the remaining Chinese labs to be measured by how fast they follow.
Background
China proposed the new export controls on October 16, covering 23 AI technologies including large models, agents and computing clusters, and has been consulting with industry on the details since the first half of the year. The rules were partly a response to Washington’s own technology restrictions — and partly a reaction to the Meta–Manus deal, in which Beijing forced the unwinding of an acquisition it saw as a strategic technology transfer. DeepSeek’s approval marks the first real-world test of the regime, and the regime passed.
Source: Reuters, The Edge Malaysia (Bloomberg)
0 Comments