ChatGPT and other generative AI tools that use large language models (LLMs) are a hot topic. Released in November 2022 by OpenAI, ChatGPT is a chatbot – it generates text output refined through user prompts.
What makes it special is how advanced and impressive that output is. The stratospheric rise of generative AI tools has sparked much debate about what this could mean for the future of education, the job market, humanity and society as a whole.
By now, you’ve probably interacted with a generative AI. But who owns the copyright on the output and how does the copyright apply?
Read more: The call to regulate AI is getting louder. But how exactly do you regulate such technology?
Text output and the law
ChatGPT is powered by an LLM – a machine-learning algorithm that processes massive datasets, including text, websites, news articles, and books. Using billions of parameters, ChatGPT statistically analyzes complex language structures and patterns to produce the output.
Some people might think that OpenAI – the company responsible for ChatGPT – would have a copyright on every output (the generated text), but that is not the case. The terms of OpenAI assign the right, title and interest in output to a user. Anyone using such AI tools should be aware of the copyright implications of output generation.
Aside from ethical and moral issues related to Academic Integritythere are many copyright implications around LLMs.
For example, if you use ChatGPT to produce output, do you own the copyright to that output under Australian law? Can AI like ChatGPT be considered a legal co-author of any LLM output? Do LLMs infringe the copyright of others by using data used to train these models?
Read more: No, the Lensa AI app isn’t technically stealing artist work, but it’s going to shake up the art world
Do you own your ChatGPT output?
Under Australian law, because the output is computer-generated code/text, it can be classified as a literary work for copyright purposes.
However, if you want to copyright ChatGPT output as a literary work, requirements known as “sustainability criteria” must also be met. When considering AI processes in light of the existence criteria, the analysis becomes challenging.
The most contested criteria of existence in the context of LLMs are those of authorship and originality. Pioneering Australian cases dictate that a literary work must come from an author.independent intellectual effort”.
To determine possible copyrights in ChatGPT output, a court would scrutinize the underlying creation processes. Hypothetically, when considering how LLMs learn, even though people turn on AI, a court would likely view this inducement as a separate, prior act to actually creating the output. The court would likely find that the output is produced by the AI. This would not meet the authorship criteria because the output is written by an AI rather than a human.
The conversation
Also, the output is unlikely to adequately express a person’s “independent intellectual effort” (another criterion of existence) because AI produces it. Such a finding would be similar to the ruling in a landmark case involving a computer generated compilation. There, a valuable Telstra database was not copyrighted due to a lack of determination of human authorship and originality.
For these reasons, copyright is unlikely to take effect on ChatGPT output as a literary work produced in Australia.
Meanwhile, under UK law, the result could be different. This is because UK law provides for a person who makes the arrangements for a computer-generated literary work to be considered an author for copyright purposes.
Can you be a joint author with ChatGPT?
In recent years, human authorship has been taken to court a few times abroad, including the famous monkey selfie business in the United States.
In Australia, a work must come from a human author, so AI is not eligible for authorship. However, if AI were ever to achieve anything akin to its own version of consciousness, debates over AI personality will raise many issues, including whether AI should be considered an author for copyright purposes.
Assuming that AI can one day be considered an author, if a court were to review joint authorship between an individual and AI, each author’s contribution would be scrutinized in detail. A “work of joint authorshipstates that each author’s contribution should not be isolated from the others. It is likely that the prompting of the AI is considered by an individual to what the AI system does next, so joint authorship would likely fail.
Do LLMs Infringe Copyright?
A final question is whether LLMs infringe the copyright of others by accessing data in training. Such data may be copyrighted material. This requires an examination of the LLM training and output. Is a substantial part of copyrighted material reproduced? Or is mass data being synthesized without substantial reproduction?
If it’s the earlier option, it may have been infringed; if it is the latter, there would be no infringement under current law. However, even if the output reproduces some copyrighted material, it may fall under a copyright exception. In Australia this is called fair dealing.
Read more: Explainer: what is ‘fair dealing’ and when can you copy without permission?
Fair dealing allows certain purposes, such as research and study. In the US, similar fair use exceptions are broader in scope, so LLM output may be affected. The European Union also has one copyright exception for text and data mining which permits the use of data to train LLMs, unless expressly prohibited by a rights holder.
Since AI is here to stay, one final point to consider is whether any changes need to be made to the Australian Copyright law to enable an AI user to be considered an author for copyright purposes. Should we change the law by following in the UK’s footsteps, or introduce a text and data mining exception similar to the one in the EU?
As AI initiatives progress, Australian copyright law is likely to grapple with these issues for years to come.
