China's State Council Lays Out IP Rules for AI and Open Source by 2030

China's New Five-Year Plan: IP Rules for AI, Data and Open Source

China's State Council has called for intellectual property protection rules for algorithms, AI-generated works and the platform economy to be developed over the next five years, as Beijing tries to keep its legal framework in step with fast-moving emerging industries. The five-year IP Protection and Utilization plan, released on July 31, also raises the possibility of protection rules for data-related IP and asks regulators to research open-source agreement rules while supporting the establishment of domestic open-source communities.

By 2030, the plan sets an ambition to strengthen China's overall IP position and global competitiveness, with stronger protection, a greater emphasis on the market value of IP, higher-quality public services, a more efficient management system, deeper international cooperation and what it calls decisive progress toward building a strong IP nation.

Xiao Youdan, a researcher at the Institutes of Science and Development under the Chinese Academy of Sciences, told Yicai that rules for AI-generated content, training data and responsibilities among platforms, model developers and users remain deeply contested. He added that open source is not simply free use but a set of conditional rules for technological collaboration, and that progress will likely come through case studies, pilot programs, industry standards and international negotiations rather than one comprehensive law.

The plan also signals a shift in how officials evaluate IP: away from raw numbers of patent applications, grants and trademark registrations and toward whether rights can be priced, licensed, financed and turned into products. According to Xiao, that change is meant to guide local governments, technology companies and research institutions toward supporting commercially valuable innovation.

From Patent Counts to Market Value: What the IP Shift Means

The Hardest Question: Who Owns AI-Generated Work?

Xiao's list of unresolved points — whether generated content qualifies for protection, who holds the rights, how training data relates to output, and how liability is split among platforms, model developers and users — explains why the State Council is setting a 2030 target rather than legislating immediately. The plan's mention of exploring data IP rules suggests Beijing sees the issue as structural, but the legal mechanics are still undefined. Companies building AI products in China therefore face a period where ownership and liability questions remain open.

Open Source as a Policy Tool — and a Compliance Risk

The cabinet's call to research open-source agreement rules and build domestic communities is not purely ideological. Xiao's point that open source is a set of conditional collaboration rules, not free use, implies that license terms carry real commercial obligations. The plan could give domestic communities formal backing while also pressuring companies to align their contributions and uses with Chinese interpretations of open-source licensing.

From Patent Counts to Market Value

The most concrete change in the plan is its evaluation logic, as Xiao describes it. Past policy rewarded the number of rights — patent applications, grants, trademark registrations. The new direction asks whether those rights can be recognized, priced and verified by the market: whether they can be turned into products, support industry value-added, enable financing, facilitate licensing transactions and compete globally. For local governments, the emphasis shifts from rewarding filings to building IP transformation platforms, financial services, price discovery mechanisms and industry application scenarios.

What the Shift Means for Companies and Universities

Xiao said technology firms should treat IP not just as a legal asset but as a tool for financing, competition, global expansion and standardization. For universities and research institutions, he urged attention to industrial prospects and transformation potential before patents are filed, rather than applying for patents solely to meet quantity targets. If implemented, that would redirect resources toward commercially viable technologies and could reduce China's patent volume in the near term.

How AI Developers, Patent Holders and Research Institutions Should Respond

For AI developers, patent holders and technology companies operating in or alongside China, the plan indicates where the rules are heading before they are written:

• Expect the first concrete rules to emerge in narrow areas first — through case studies, pilot programs and industry standards — rather than a comprehensive IP overhaul. Follow regulator guidance linked to the plan on AI-generated content and data, which Xiao identified as the most contested areas.
• Review who holds rights and liabilities across your AI value chain now. The unresolved questions over training data, output ownership and the split of responsibility among platforms, model developers and users are likely to become contractual and regulatory battlegrounds.
• Treat open-source participation as a business decision, not a default. Xiao's emphasis that open source is conditional collaboration means license terms can impose commercial compliance obligations on contributors and users.
• Reorient IP strategy toward market value. The plan explicitly shifts evaluation from patent counts to pricing, licensing, financing and productization, so portfolios that cannot be monetized may receive less government support.
• For universities and research institutions, design IP strategy around industrial application and transformation potential before filing, rather than quantity-based incentives that the plan is moving away from.