“A computer – even if it is artificial intelligence – does not qualify as an author”

Enlarge

Matthias Leistner is Chair of Private Law and Intellectual Property Law, with Information and IT Law.

“The possibilities and limits of artificial intelligence (AI) have been the subject of intense discussion for at least a decade in expert circles — including in legal circles with regard to private law and copyright issues. The prevailing view is that it is important not to throw the baby out with the bathwater and also not to place too many exaggerated hopes in the current performance of the algorithms, even given what appear to be some surprising achievements.

The capabilities of today’s large generative AI models are impressive. They are undoubtedly generated in an unpredictable and partly independent way. However, for many reasons, there’s no way this can be called real “intelligence”.

Meanwhile, the hype around AI has also reached the general public. Right now, it is ChatGPT that sparks the imagination of columnists, business editors – and also some lawyers. In this context, the question of copyright regarding such software applications is obviously of interest now.

Who has ownership of texts produced by ChatGPT?

From a copyright perspective, there are several aspects to this question. First, there is the issue of who has ownership of texts produced by ChatGPT. The answer is similar, with certain nuances, under currently applicable law across virtually all major jurisdictions worldwide: There must be a human author. A computer — even if it uses artificial intelligence — does not qualify as an author (in the sense of copyright law).

Of course, humans can become authors through their control of the algorithms or through personal selection and post-processing of texts that are ultimately published. But merely formulating a query or a task for an AI application like ChatGPT to complete is not sufficient for this.

The fact that AI thus currently does not enjoy any protection under intellectual property law for its output immediately prompted calls by some scholars for new intellectual property rights — be they copyrights or more limited related rights. However, these calls are premature and probably even unjustified in the long run. The very fact that such applications are already emerging in the markets on a broad scale and are evolving rapidly proves that there is currently no real reason to offer any additional incentives by granting new intellectual property rights. But the discussion is only just beginning. We need to observe the further development to see whether a need for protection may in the future arise in very specific areas — the generation of valuable training data for algorithms is repeatedly mentioned in this context, for example — that requires a legislative response. But that is all still a long way off. We should certainly not rush into any regulatory initiatives in the field of intellectual property law.

Therefore, as things stand, neither the ChatGPT application nor its programmer have any legal rights to the results generated by ChatGPT. Generally, the users have no copyright either, unless they contribute their own materials on a large scale and thus control the AI in a very specific way so that the original materials are specifically reflected and still identifiable in the results. So far so clear.

What about third party copyrights?

The more substantial copyright problem lies elsewhere: As a user, can I rely on the texts produced by an AI being copyright-free in regard of the source materials the algorithms used? This is where users must proceed with great caution.

The underlying software undoubtedly makes more or less direct use of source materials scraped from the internet to optimize the probabilistic algorithmic models underlying ChatGPT. The developers claim that all materials used in this respect were freely available on the internet and are therefore copyright-free. However, things are not necessarily that simple — especially in German and continental European copyright law.

Information as such is indeed generally considered to be copyright-free. But what happens if there are parts of texts or text structures that were created by a human author, made available on the internet and then taken and reused on a larger scale by the AI in a different context?

Under European law at least, the mere fact that texts are made available on the internet does not automatically mean that they may also be reused extensively for training or other use in AI applications.

Therefore, if the text fragments used in this respect are extensive enough to be protected by copyright (this could potentially be the case for one or two particularly original sentences), ChatGPT’s ‘product’ could still infringe the copyrights of third parties. The user, i.e. the person using the software, will then be jointly liable if they use the produced texts in a way that falls under copyright law.

Caution is therefore advised if texts produced by ChatGPT are to be copied, published or otherwise used outside the private sphere in a way that falls under copyright law. Currently, one cannot rely on the results being guaranteed copyright-free in this context. There is no real clarity over whether only truly free materials from the internet were used, or whether the text generated by ChatGPT might contain text fragments that are themselves protected by third party copyrights.

Limitations for text and data mining

Against this background, from a scholarly and legal policy perspective, a number of completely different, more fundamental questions arise. How do we ensure that the necessary source materials for the dynamic, innovative development of artificial intelligence, in particular, training data of all kinds which are essentially the bottleneck here, are available in a form that is sufficient, easily accessible, and free of charge wherever possible – at the same time taking into due account the justified interests of copyright holders?

Exceptions to copyright protection do exist or have been specifically provided for in the main copyright systems worldwide: the very flexible, rather broad ‘fair use’ doctrine in the United States; the new limitation for text and data mining in the second major copyright directive of 2019 in Europe; and broad limitations to copyright in Japan that make the country a veritable ‘copyright eldorado’ for new technologies.

The EU exception is fairly modest compared to those in some other countries, namely the U.S. or Japan. In particular, it does not allow unrestricted text and data mining for commercial purposes. It is here, where the EU might eventually find itself at a competitive disadvantage in the development of AI, not around the trite demand for additional copyright protection for the products of artificial intelligence: Thus it might be necessary to take a closer look at the legislation in regard to copyright exceptions and access in the future and possibly come up with intelligent, balanced instruments to further simplify access to copyrighted materials for this specific purpose.

At present, however, the most important thing is not to rush into any additional copyright regulation. Here, we can certainly learn from the U.S. approach: It is precisely the most innovative, dynamic solutions and applications that often develop best when they are given a certain amount of freedom at the beginning of their development. If specific problems then arise in the markets, it is generally still possible to react with legislation in good time. From the view of experts, ChatGPT is – for the time being – primarily a hype. Further research and a calm hand are necessary before any demands for changes in copyright law should be made here."