China’s Artificial Intelligence Law (Draft of Scholars’ Suggestions) was released on Saturday, following an earlier draft by scholars from the Chinese Academy of Social Sciences published in August. The draft favors developing the artificial intelligence industry over protections for users of AI systems and other humans.
It is in part a responsive to US-China AI competition: according to co-author and professor at the China University of Political Science and Law Zhang Linghan, amid today’s fierce global competition, “Not developing is the greatest security risk.”
Despite the focus on industrial advancement, providers would, per the draft, have to adhere to China’s existing data and personal information protection laws. They would also be subject to some curbs on data use. For example, they “should not use AI to analyze and evaluate individuals’ behaviors and habits, interests and hobbies, or their economic, health, and credit information.”
But in general, the draft is focused on creating a legal structure that benefits the rapid development of the industry. It highlights contributions the state should make to the AI industry, including policy support such as tax incentives and cultivating talent in universities.
And the law engages with how and whether AI providers can appropriately use copyrighted material for training purposes, coming out in favor of developing the industry over protecting human work. According to Article 24, AI providers can use copyrighted data in training sets “if the use behavior is different from the original purpose or function of the data” and “does not unreasonably harm the legitimate rights and interests of the data rights holder.”
While those stipulations do leave room for some legal maneuvering—and the draft adds that AI providers are responsible for “conspicuously” labeling the source of the data—the partiality to model development is hard to miss. It also to some degree opposes last month’s Guangzhou Internet Court ruling that stated an AI provider had misused copyrighted material in its training data.
In China, like everywhere else, the issue of how and when to protect human creativity and support the strategically essential AI industry forward is the subject of disagreement among and beyond legal experts. The authors behind the new draft law come out strongly in favor of industrial development, even as the law (in concert with other laws and regulations already on the books) would provide some user protections – just not as many as some users and legal professionals may have preferred.
In another pro-industry provision, the draft law favors copyright protections for AI-generated outputs “according to the degree of the user’s contribution to the final presentation of the content,” as long as the subject of the copyright or patent application is a human being. AI programs can’t modify AI-generated content and apply for copyright protections, but humans using text or image generators can—as long as they sufficiently alter the outputs. This is in line with the Beijing Internet Court’s November decision that ruled a user could be granted copyright protections for work created using an AI image generator, since the user had sufficiently altered the output.
AI regulation and development is transnational by nature. The draft law can be read as Chinese scholars’ acknowledgement that Beijing’s time to contribute to international AI norms is limited—and those contributions will only be taken seriously if China is considered an AI power.
China continues to pitch itself as the developing world’s leader on AI through the Global AI Initiative and other diplomatic appeals. Its AI law, when passed, could have far-reaching consequences. Once again, it would be far ahead of the United States on laws and regulations. Given the draft law’s focus, the plan is to also, eventually, dominate development.
