The relationship between ChatGPT and intellectual property is at the center of the debate. Every AI tool, such as ChatGPT, Gemini, Dall-E, and xAI, has its own terms and conditions, but there are legal gray areas around using AI-generated content as your own.
IP: what is copyright?
Copyright is a type of intellectual property that grants authors rights and powers over their works simply because they created them.
Argentina’s Intellectual Property Law (Copyright Law No. 11,723) considers an author to be the natural, human person who creates a scientific, literary, or artistic work. In other words, AI would not fit into this category.
Is LLM output protected by copyright?
ChatGPT and intellectual property: friends with benefits (and rights)
According to ChatGPT’s content policy and terms of use, users own everything they create through the LLM, including text and images.
LLM stands for large language model, a model that can understand and generate human-like text. The technical definition is:
An LLM is a statistical model that determines the probability of a sequence of words occurring in a sentence.
GPT is one of these LLMs. It uses neural networks trained on vast amounts of text found on the internet, including books, newspapers, forums, recipes, scientific papers, patents, encyclopedias, and more.
In other words, users are free to do whatever they wish with what they generate with the AI, unless a specific term or condition says otherwise, and regardless of whether they use a free or a paid plan.
That is because ChatGPT tries to avoid outright plagiarism in the content it generates. Here is an example of how it does so.
We asked ChatGPT (free plan): “Can you quote the first page of a book by Charles Dickens?” When it did not give us the specific text, we asked why.
We can see how ChatGPT itself explains its decision, which is related to copyright and intellectual property. It is therefore very unlikely to reproduce a protected work verbatim.
If you are seriously considering protecting the copyright in content generated with ChatGPT, you should look into the intellectual property laws of your country.
For example, in the United States, the U.S. Copyright Office does not allow copyright in machine-generated content.
In the United Kingdom, by contrast, the Copyright, Designs and Patents Act provides that computer-generated works can be protected by copyright. It all depends on where you are and which bodies regulate the matter.
Where does ChatGPT get its data?
Many users simplify their tasks and creative work with this tool, but the truth is that much of the content it generates is not reliable. Why? Because its answers include a mix of copyrighted works.
Here is what ChatGPT itself told us:
OpenAI’s service policies
We already know that content generated by ChatGPT is not protected by copyright. OpenAI’s usage policy makes it clear that using the generated output in illegal or harmful ways is prohibited, and it encourages responsible and safe use of AI.
OpenAI’s usage rules:
- Legal compliance: respect the laws of each region and the privacy of others, and avoid harmful or illegal activities, including putting minors at risk. Do not use ChatGPT to facilitate harmful or illegal actions.
- Your safety and that of others: do not use the services to promote harm, develop weapons, or injure others.
- Integrity toward others: do not use the generated information to defraud, spam, intimidate, harass, or discriminate.
- Respect for safeguards: do not circumvent OpenAI’s safety measures, except in cases of authorized research.
Conclusion
Whether AI-generated content can be registered for copyright depends entirely on the country you are in. As we understand it, ChatGPT does use copyrighted content to train and to provide information, so we must be careful and review what it produces.
Responsible, appropriate use of AI can bring us many benefits, but to get there we first need to understand what we are doing and what we want from these tools. That way we will know how to use them without running into trouble with anyone’s intellectual property.
And note ChatGPT’s last answer. After the conversation we started, this artificial intelligence acknowledges (something it did not do at first) that it uses copyright-protected content. Be very careful with what ChatGPT creates and we take as our own. It could get us into trouble.
Related reading:
