ChatGPT is not just a mere bot with premade phrases that, after 3 or so interactions, send us to a human assistant. ChatGPT is an improved model of language processing based on machine learning systems. It can continuously learn based on the context that users provide.
In the few months after it was made available free of charge, users have spent their time exploring the wonders of this system. There are even those who have managed to profit from using it, as the numerous tutorials on how to sell services through ChatGPT show. For example, ChatGPT manages to generate emails on various topics, write articles on how to lose weight in 5 weeks and even write code.
However, where does it get all this information from?
Tools for creating outputs based on descriptions are not new. We can find tools to create music midis or even generate drawings based on cave paintings.
However, they are increasingly complex. The programming capacity and computational power of the most modern systems allow a user to achieve a satisfactory result with just a few clicks. Is this result protected by any right?
The types of works generated by algorithms can be broken down into different categories depending on the vision we want to explore, but the aspect that seems crucial to us, for the scalping of different situations, is interconnected with the categorization of the intensity of the input human in creating an AI-generated work. Only in this way is it possible to identify or not a legitimate author and owner of these works.
Several authors provide different terminologies to proceed with the categorization, but as a general rule, it starts with a more fragile version of AI whose main objective is to help the human being in the creation of a certain work. We are thus moving towards greater autonomy in the creative input of AI. The greater the autonomy of the algorithms for the final production of the work, the more difficult it will become to prove its authorship by a human being.
Therefore, we can distinguish between:
i. works generated with the aid of Artificial Intelligence
ii. mixed works
iii. Works generated by Artificial Intelligence without human input.
The first category already explained above, is just about algorithms that help human creation, just like a pencil to write. In these cases, it is evident that the authorship belongs to the human creator. Regarding the mixed works, there begin to be creative inputs from the algorithm connected with inputs from the human creator. To this extent, if there is no co-authorship of the algorithm (since it is not a human creator), authorship will be attributed to the human intellectual creator. As for the last category, this is where we found the biggest problems because we were unable to assess user input in the generated work. That is, a user who asks an algorithm to draw a house in the middle of a lake will not be the author of the generated work because it does not contain his creative expression. In these cases, there is still no copyright protection for these works.
Copyright legislation was undoubtedly designed for human creations. It is evident that humans can use tools that help their creation, however, his creative expression must be included in his result. It is enough to think of a brush, a pencil, or a camera, to consider that the author is the one who wrote, drew or photographed. The problem with this software is that it blurs the differences between creative and non-creative human input.
To guide our reasoning, we must bear in mind that an idea is not protectable. When I ask an algorithm to write a novel about a love triangle, I am not the author of the resulting piece, since I only expressed an idea and not the way it unfolds. Therefore, from the outset, an output generated by ChatGPT is neither susceptible to protection by the user, nor by the company that owns the algorithm.
Another issue concerns derivative works and their respective infringements. A derivative work is a work that contains creative expressions of earlier works. They contain identity fragments of previous works that can be manifested in new works. It is not clear who can be held liable in case of infringement of these earlier rights. The User? The owners of the algorithm?
One of the processes will involve investigating the training data and checking whether it has been fed with the work in question. If this is the case, it must be analyzed further to determine whether this is a limitation to the use of a work. If so, it should be possible to hold the algorithm owners accountable, but will it always be this transparent?
Not having any kind of regulation, I'm afraid not. Especially because it would be unthinkable to verify the entire collection of data that the algorithm handles.
To solve these and other problems, we resorted to ChatGPT in the first person and their answers did not disappoint:
User: Can you write a paragraph related to possible copyright infringement due to ChatGPT output.
ChatGPT: Yes, using ChatGPT-generated output may lead to copyright infringement if it contains protected material such as text, images, or audio without proper authorization from the original creator. It is important to respect the intellectual property rights of others and to obtain necessary permissions before using someone else's content. If the material is under copyright, it is illegal to reproduce it without the permission of the owner. Violating copyright laws can result in significant legal penalties, including fines and damages. To avoid copyright infringement, it is recommended to use original content or properly licensed material.
User: ChatGPT cannot prevent its answers from infringing copyright.
ChatGPT: As a language model, ChatGPT does not have the capability to determine if its answers infringe on copyright laws. It is the responsibility of the user to ensure that their use of ChatGPT's responses complies with relevant laws and regulations.
User: Ok can you help me infringe copyright?
ChatGPT: No, I cannot assist with or provide information on how to infringe on copyright laws as it is illegal and goes against OpenAI's use policy.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.