Using ChatGPT to create editorial content can be an interesting way for companies to produce relevant content quickly and effectively. However, there are some legal pitfalls to be aware of when using ChatGPT.
What is ChatGPT?
ChatGPT is an artificial intelligence developed by the US company OpenAI that is able to conduct human-like conversations. It was originally developed for use in chatbots and virtual assistants, but has now also proven itself in other areas, for example as a tool for creating social media content or automating customer inquiries.
One of the interesting areas of application of ChatGPT is the creation of texts that are perceived as genuine by humans. This can be used, for example, to create artificial blog posts that appear so authentic that they are not recognized as such by human readers.
It should be noted that ChatGPT (currently) has a closed knowledge base from the end of 2021 and therefore cannot access current events. So if you ask ChatGBT, for example for example, about Russia’s attack on Ukraine, the program explains the annexation of Crimea in 2014.
Artificial intelligence and copyright law
But what does this mean for copyright? Can ChatGPT really be regarded as the author of a text and is he/she/it therefore entitled to copyright?
According to Section 1 of the German Copyright Act, any creative work belonging to literature, science or art enjoys copyright protection. This also includes texts created by artificial intelligence. However, there is an important restriction: copyright only applies if the creative work actually originates from a human being, as an AI is not considered a creator within the meaning of Section 7 of the Copyright Act. So if you give the program the command to freely create a text, for example a fairy tale, ChatGPT is not the author of the fairy tale due to the lack of human creation. However, the user giving the input command cannot be regarded as the author either, as he or she has not provided the copyright-relevant service, the linguistic and artistic creation of the fairy tale, himself or herself. The result is a copyright-free space – an intolerable legal situation that urgently requires a sensible solution.
ChatGPT may infringe the copyright of others
However, there are some pitfalls that should be considered. If the text was merely “reworded” by ChatGPT and does not contain any significant changes or additions, the copyright of the original author could be infringed. In this case, it is advisable to obtain the author’s permission or use other sources.
Particularly problematic is the fact that the language model does not cite any sources, so identifying an author other than ChatGPT is likely to prove difficult. When asked about the sources of a text created by ChatGPT, the program regularly replies: “I am a language model and have no way of citing or linking to sources.”
One way to avoid this is to use ChatGPT only as a source of ideas or inspiration and not as a direct source for the text. In this way, you can ensure that the text is your own original creative work and not a copy of existing material.
Companies should therefore ensure in advance that they have the necessary authorizations to use ChatGPT texts in order to avoid legal problems. OpenAI has the right to approve or reject the use of ChatGPT-generated content, depending on whether the text is suitable for the intended use or whether there are other restrictions.
Companies should therefore contact OpenAI in advance and ensure that they have the necessary rights to use the ChatGPT-generated posts. In this way, they can ensure that they are acting lawfully.
It is also important that the person who created the text takes the necessary precautions to ensure that the text does not contain false or misleading information. Sometimes ChatGPT simply invents facts, but then presents them as highly plausible. This can lead to copyright issues if the text is used by others and causes damage.
Here is a brief example from our law firm practice: We asked the chat program whether it knew a specific paragraph from the Copyright Act. This was confirmed. We then asked about current case law on the subject. The program immediately listed three current judgments, including the case numbers and headnotes of the judgments. After an Internet search and subsequent inquiries with the relevant courts, it turned out that these judgments had never been made. However, the file number was basically correct and the headnotes of the judgments were not unreliable. This shows how dangerous the use of ChatGPT can be, especially for scientific work.
Great potential
Overall, ChatGPT offers many options for creating authentic texts, as long as the copyright of others is respected and the text does not contain any false or misleading information. However, ChatGPT is a useful tool to make your work easier, especially as food for thought or for the use of text modules. ChatGPT is also suitable for a first draft, which is then thoroughly checked, revised and edited, even if any copyrights are respected.
Ultimately, however, it should be borne in mind that the AI is currently still a test version. The licensing of use and the application of a sophisticated general terms and conditions system are foreseeable for the future. It remains to be seen what the rights of use will look like.
For the authorship of artificially generated images, read our article on AI image generators.
Focus on AI & copyright
In our thematic focus “AI & Copyright“, we deal with the copyright aspects of current developments in the field of artificial intelligence. So far, this series has included introductions to the applications ChatGPT and the use of AI art, an overview of the terms of use and training data of popular AI image generators and the Mandelbrot debate in the context of the legal discussion. We shed light on the concerns and fears of press publishers in the article“Ancillary copyright and AI” and explore the fundamental question of whether AI can be an author in a separate article.