Generative AI tools like ChatGPT, which rely on large language models (LLMs), have rapidly gained attention since ChatGPT’s launch in November 2022 by OpenAI. While users marvel at the sophistication of its output, questions about its broader implications for society and the legal frameworks governing it—particularly copyright law—are becoming increasingly relevant.
Understanding ChatGPT and Its Outputs
ChatGPT, powered by a machine-learning algorithm trained on vast datasets, generates text by analysing complex language patterns. While OpenAI provides the platform, its terms assign ownership of generated outputs to users. But does this mean users hold the copyright to the outputs?
This question is far from straightforward and requires careful legal analysis.
Who Owns ChatGPT’s Output?
In Australia, copyright law classifies computer-generated text as a “literary work,” provided it satisfies key criteria, such as authorship and originality. Under Australian precedent, a work must reflect independent intellectual effort to qualify for copyright protection.
The challenge lies in whether prompting an AI like ChatGPT constitutes sufficient intellectual effort. Courts would likely view the user’s role as a preliminary act, with the actual “creation” attributed to the AI. Because the output originates from a machine rather than a human, it fails the requirement for human authorship, making copyright unlikely under Australian law.
Interestingly, UK law provides a contrasting approach, allowing the person who arranges for a computer-generated work to be considered the author. This divergence highlights the complexities of reconciling traditional copyright principles with modern technology.
It also creates complexity about the physical location of where someone is when they access generative AI to determine if the human involved has ownership of the creation. What if someone is on a plane flying from the UK to Australia and make various prompts to ChatGPT in both locations?
Joint Authorship with AI: Is It Possible?
In Australia, authorship must originate from a human, ruling out AI as a potential author. Even in hypothetical scenarios where AI achieves sentience, courts would need to determine whether a user’s input and the AI’s processes are sufficiently intertwined to constitute joint authorship. Current analysis suggests the separation between human prompting and AI processing would likely preclude this possibility.
However, debates around AI’s potential sentience and personhood could reshape this area of law in the future. If AI tools were recognized as legal entities, questions of joint authorship, intellectual property rights, and ethical considerations would become increasingly complex. Legal scholars and practitioners must closely monitor these developments to anticipate how such recognition could impact existing frameworks.
Is Copyright breached by LLMs?
Another key issue is whether LLMs infringe copyright by using protected material in training datasets. If substantial portions of copyright-protected works are reproduced, infringement could occur. However, synthesizing data without directly replicating substantial portions may avoid infringement.
Copyright exceptions, such as Australia’s “fair dealing” provisions or the broader “fair use” in the US, can also play a role. For instance, the European Union’s exception for text and data mining permits the use of protected material unless explicitly prohibited by the rights-holder. As AI continues to advance, the definition of “fair use” and the boundaries of copyright protection may require reinterpretation or reform.
The Future of AI and Copyright in Australia
As AI tools like ChatGPT become entrenched in society, Australian copyright law faces significant challenges. Should the Copyright Act be amended to recognize AI-generated works? Could a framework similar to the UK’s or the EU’s offer a balanced approach?
These are critical questions for legislators, legal professionals, and businesses alike. With AI’s rapid evolution, legal systems must remain agile to address emerging issues. Policymakers must strike a balance between encouraging innovation and protecting intellectual property rights. Collaborative discussions involving legal experts, technologists, and policymakers will be essential in shaping a future-proof legal framework.
As generative AI continues to evolve, so too must the legal frameworks that govern its use. The ongoing discourse on AI and copyright will undoubtedly influence the development of intellectual property law for decades to come.
About the Author: This article has been authored by Steven Brown. Steven is a Perth lawyer and director and has over 20 years’ experience in legal practice and practices in commercial law, dispute resolution and estate planning.
