Artificial intelligence (AI), has brought great convenience to the human world, but has also cast a great challenge on existing rules, especially on copyright laws in all jurisdictions. It is controversial, for both the scholars and the practitioners, whether an AI-generated work can be protected by copyright law. Recently, the Beijing Internet Court, for the first time in China, tackled the relevant issues of a software-generated report in Beijing Film Law Firm v Beijing Baidu Network Information Technology Co., Ltd. (the "Film v Baidu case"). The ruling in this case is likely to be of great significance for future AI-related cases.
Wolters Kluwer China Law & Reference ("Wolters Kluwer"), a legal database, with the search conditions set by the claimant, Film Law Firm, produced a report based on the data of cases of the film and entertainment industry (referred to as the "Report"). Then, the claimant created an article entitled “Analytical Report on the Judicial Big Data in the Film and Entertainment industry: Film Industry in Beijing” (the "Article") based on the Report and published it. Without the claimant's permission, the defendant deleted the claimant's name (which was an indication of authorship), the foreword and the research introduction part of the Article and also published it on the internet.
The claimant sued the defendant and claimed that the defendant infringed its copyright, including the right of authorship, the right to protect the integrity of its works and the right of dissemination through the internet. However, the defendant counterclaimed that the Article, including the words and the charts within it, was automatically produced by a legal database software, not created by the claimant, and that the data was not collected by the claimant either. Therefore, the Article was not created by “human intelligence”, and not under the scope of protection of the Copyright Law.
Result: The Beijing Internet Court held that the charts in the Article did not constitute a “graphic work” because of lack of originality, but the words in the Article, had originality and constituted the claimant's “literal work”. However, the Report generated by Wolters Kluwer by the "visualization" function did not constitute a work under the Copyright Law.
Analysis of the court's decision
This case focused on the following two issues:
- Whether the report generated by the software constituted a “work” that was protected by the Copyright Law?
- Whether there was someone or some entity who could claim rights to the report generated by the software? If yes, what were the rights?
The court answered the first question as follows:
a) The report generated by the software did not constitute a “work” that was protected by Copyright Law
The court did rule that the defendant infringed the claimant’s copyright, but found that the claimant copyright lay in the Article (created by the claimant based on the software-generated Report), instead of the Report itself. In fact, the court specifically pointed out that the Report did not constitute a “work” that is protected by Copyright Law. The reasons are as follows:
I) The charts in the Report did not have originality, but the words did
Article 2 of the Regulations for the Implementation of the Copyright Law stipulates that "The ‘works’ in the Copyright Law refers to intellectual achievements that are original in the fields of literature, art and science and can be reproduced in certain tangible form." Therefore, an intellectual achievement should be of originality, in order to constitute a “work” in the Copyright Law.
The court conducted an examination during the trial, following the same steps as the claimant did when they used the Wolters Kluwer, and produced a report (the "Examined Report”). The court pointed out that the differences existing between the charts in the Report and the Examined Report were simply from the changes in the database, instead of intellectual creation. In other words, the charts would be exactly the same as long as the two reports were based on the same data. Therefore, the court held that the charts in the Report were not “works” in the Copyright Law because of the lack of originality.
The claimant also argued that the charts in the Article had been beautified based on those in the Report. The court rejected this argument on the basis of a lack of evidence showing that the charts that have undergone some beautification, such as adding some frames or colors, which might have resulted in it being protected by the Copyright Law.
In summary, the charts produced by computer software had no originality because their contents were totally dictated by the data. But, the charts with beautification, if that reflects the author’s original creation, could constitute a “work” and be protected by Copyright Law. When considering whether one picture is similar with another, the expression of the pictures automatically produced by the software should be eliminated while only focusing on the results of the further beautification process.
The court held that the words in the Report had originality. The words, different from the charts, were the result of selecting, judging and analysing the data. The analysis, from various perspectives, varied even based on the same data and charts.
II) A copyright work should be created by humans
Although the court concluded that the words in the Report had originality, it still denied them as a copyright work. Article 9 of the Copyright Law stipulates that: “Copyright owners shall include: (1) Authors; and (2) Other citizens, legal persons, and other organizations that hold copyright under this Law.” Therefore, under Chinese law, copyright owners should be natural persons, legal persons or other organisations and the copyright of the latter two also comes from a transfer of the natural persons who create the work. As the court said in the decision, “being created by human is still a necessity for a copyrightable work”.
In this case, the contribution of the software developers and the users into the Report produced by Wolters Kluwer did not include their original expression. The developers provided the software programs and the users input the research keywords, neither of which constituted the “creation” process of the Report. Therefore, neither the developers, nor the users were the creators. In fact, it was the computer software, not the human beings that “created” the Report. Consequently, the court concluded that the Report was not a “work” under the Copyright Law.
However, the Article was created by the claimant through further editing, reorganizing and analysing the Report, which brought a materially big difference into the Article, compared to the Report. The difference reflected the claimant’s intellectual activities and constituted a “creation” process. Therefore, the court held that the Article was a “legal person’s work” created by the claimant, judging from the evidence submitted by the claimant (without evidence to the contrary.)
Second question: software users are entitled to claim rights in the software-generated Report
Although the software-generated Report was not a copyrightable work, the court said that it was still protected by law. Since the Report contained the contribution from the software-developers and users and enjoyed the dissemination value, rights should be given to the contributors. Assessing the balance of interests, the court considered that the rights should be given to the software users. The developers' returns on their investment had been realised through collecting the royalties from the users and they were not motivated to distribute the reports generated by the software. However, the users obtained the reports by paying the royalties and were motivated to further use and distribute the reports. Therefore, the court concluded that the software users should be entitled to the rights in order to promote the culture communication and prosperity.
It should be noted that the court did not clarify what kind of rights existed other than copyright. The authors of this blog think that the neighbouring rights can be referred to under this scenario.
Conclusion and comments
Film v Baidu is the first case in China for the courts to clarify controversial issues over AI-generated works such as the nature and type of rights. The court decision clearly stated that under current law systems, the charts in the Report are directly determined by the data, and therefore not original, while the words, although original, could also not qualify as a copyright work because they were not created by a natural person.
The court accepted that some sort of protection should be available from the software-generated work to motivate the investment and that this should be awarded to the software users. This raises further questions, including:
- What right specifically is it, since it is not copyright? As mentioned above, we think that neighbouring rights are applicable under this scenario, with the scope of the rights to be specifically stipulated.
- If the works generated by the software do communicate the motions and ideas of the developers or the users, is it possible that they can be entitled to the copyright of the works? Does such software “create” the work, or it simply produce as a tool?
We leave these questions open for further discussion and welcome any feedback from readers.
In this context see also these recent blogs by our UK colleagues: AI-generated inventions: no longer science fiction and From machines to plants: the battle continues over non-human inventions.
Xu Lin Partner
(T): +86 21 3122 2068
(M): +86 138 1898 1516
Suzie Li Paralegal
(T): +86 21 3122 2068
(M): +86 139 1671 2191
Yu He Intern
(T): +86 21 3122 2068
(M): +86 182 0509 5702
联系方式：021-3122 2068/138 1898 1516
联系方式：021-3122 2068/139 1671 2191
联系方式 : 021-3122 2068/182 0509 5702