How ChatGPT is changing copyright law: a glimpse into the future

One of the paragraphs below was generated by ChatGPT. It is up to you, dear reader, to determine which one it is. You can email Niels van den Bogaard's answer and explanation via the contact form below. The one with the sharpest eye, in the author's opinion, will be sent a delicious bottle of wine as the winner of this Turing test.

Date: Feb. 28, 2023

Modified November 14, 2023

Written by: Niels van den Bogaard

Reading time: +/- 2 minutes

One of the paragraphs below was generated by ChatGPT. It is up to you, dear reader, to determine which one it is. You can email Niels van den Bogaard's answer and explanation via the contact form below. The one with the sharpest eye, in the author's opinion, will be sent a delicious bottle of wine as the winner of this Turing test.

The bottle of wine has already been given away, but if you are curious if you know the right answer you can always leave a message.

ChatGPT and copyright

In my previous blog, I provided some background on copyright in packaging and labeling. Copyright is the creator's exclusive right to reproduce or disclose a work. This right, of course, is broader than just resting on packaging and labeling. It can also rest on texts, such as this blog. But what about copyright in texts created through a text generation system, such as ChatGPT?

How does ChatGPT work?

In the case of texts from ChatGPT, texts are created based on questions asked by the user. The GPT-3 language model behind ChatGPT, is trained on many billions of text examples. Its purpose is to learn the most logical next word - in short - based on a prompt and the words already given. The text examples are gathered on the Internet by a "crawler" (software that automatically scours the Internet and collects relevant data), which makes the online information useful to the text generation system.

Fair use

It is important to emphasize that ChatGPT does not have unique thoughts or ideas, but rather a collection of information learned from a large amount of text on the Internet. This means that the texts generated by the model often already exist in one form or another.
In this sense, the system makes use of "fair use" of public or non-public data.

Resources

I deliberately say public or not, because it is difficult to trace the source of the texts generated by ChatGPT. After all, this is a dataset composed of more than a trillion (one one with twelve zeros) words. This makes it difficult for authors of texts used by ChatGPT to prove that they are copyright owners of (parts of) the texts generated by ChatGPT. Unless it is a word-for-word quote. I haven't come across an example of that yet.

Does ChatGPT violate copyright law?

Whereas it will be difficult for the copyright owner of the source texts to prove that their text was used for ChatGPT, the same is true for texts generated by ChatGPT. Not for nothing do students eagerly use the system to generate pieces of work that are indistinguishable from real ones. Although such a text is generated based on existing works, it may be an original work within the meaning of the Copyright Act with a copyright vested in its creator. The unusual thing then, is that the originator of the text in question is basically computer software, namely the ChatGPT program.

ChatGPT as a creator?

In such a case, one can reason that the creator of ChatGPT has to be regarded as the creator, where ChatGPT can be regarded as the indirect creator. An analogy can be made, for example, with the copyright on works created by employees, for which the Copyright Act provides in Article 7 that the copyrights thereon in principle belong to the employer.

Change in legislation?

On the other hand, there are social noises that certain algorithms should be given independent positions (see, among other things, the AI Act being developed on behalf of the European Commission), especially where autonomous systems for vehicles, weapon systems, medical assessments and the like are concerned. It is possible that ChatGPT itself will eventually become the bearer of Copyrights and such software will move in the direction of legal entities. Human-created entities that can be independent carriers of rights and obligations.

Right now we are in the middle of this discussion. Interesting times without question. And if you have any questions about copyright, don't hesitate to contact me or my colleagues. We will be happy to help you.


Stay Focused

As attorneys for business owners , we understand the importance of staying ahead. Together with us, you will have all the opportunities and risks in sight. Feel free to contact us and get personalized information about our services.