Can Artificial Intelligence Be an Inventor? The Ongoing Debate in Australia’s Patent System

Written by Barry Meskin on 14 November 2025

In today’s world of generative AI, machine learning and autonomous systems, one of the hottest legal questions is this: can an AI be listed as an inventor on a patent? The idea sounds futuristic, but it has already played out in courtrooms, legislatures and patent offices. In Australia, that debate has sharpened significantly.

Let’s explore how the law views AI inventorship, the key cases that have shaped the rule and how IP Guardian helps clients navigate this uncharted territory.

Why the question matters

Imagine a powerful algorithm that independently conceives a novel drug, a new battery material, or an optimised mechanical structure without explicit human direction. If patent law allows AI to be inventor, it would radically shake how we reward creativity, assign ownership and drive investment.

Patents are incentives. They reward those who make the inventive leap. If AI starts making those leaps, the whole system must rethink who gets credit, who pays and who enforces rights. That makes this more than academic. It is foundational to technology policy, innovation strategy and business planning.

The foundational: inventors must be people

Under Australian law, the concept of an inventor is rooted in the idea of a person who devises or contributes to a new invention. The Patents Act 1990 (Cth) and relevant regulations have long assumed that inventors are natural persons. 

IP Australia’s manuals make this clear: “an inventor must be a natural person and cannot be a legal recognised entity such as a corporation, or Artificial intelligence.”.

That assumption was tested in the famous Thaler / DABUS saga. Dr Stephen Thaler applied for patents naming an AI system called DABUS as the inventor, claiming the AI had autonomously generated the invention. The Deputy Commissioner initially rejected the application for failing to name a human inventor.

Some stunned observers looked on with interest when, in 2021, Justice Beach, a Federal Court judge, found that the term “inventor” might be interpreted more flexibly, potentially allowing a machine to be inventor under certain readings of the statute. But that was not the end.

In 2022, the Full Federal Court unanimously reversed that view in Commissioner of Patents v Thaler, reaffirming that an inventor must be a human (or group of humans). The court emphasised that in Australia’s statutory structure, invention attribution must be to persons.

Dr Thaler sought appeal to the High Court, but his application for special leave was refused in November 2022. That refusal effectively locked in that AI cannot be an inventor under current Australian law.

Thus, the prevailing rule in Australia today is: you cannot list an AI as the inventor.

(Image: Thaler / DABUS case ruled inventor must be human)
(Image: Thaler / DABUS case ruled inventor must be human)

Global contrasts spur ongoing debate

Australia is not alone in wrestling with this question. Many jurisdictions maintain the stance that only humans can be inventors, including the U.S., the European Patent Office and the UK.

Some intrigue surrounds jurisdictions like South Africa where patent offices granted rights listing AI as inventor (with human beneficiaries). Yet even in those cases, thorny issues of ownership, liability and accountability loom large.

In Australia, one interesting nuance is that while the inventor must be human, inventions that result from AI use or AI assistance can still be patentable, provided a human played a decisive role in conception or implementation. The courts and IP Australia require clarity about human “inventive concept” contributions.

The rise of generative AI, large language models and self-optimising systems stirs policy discussions about how IP law might adapt. Some observers look to the Artificial Inventor Project, a campaign pushing test cases in multiple countries to force law reform to recognise nonhuman inventors.

Challenges and the tensions of policy

Several deep challenges lie at the heart of the debate:

  • Attribution and accountability: A patent grants rights and obligations. If a machine is inventor, who owns those rights? Who is liable if the application is wrong or infringing?
  • Incentives and reward: The patent system rewards those who take the risk to invent. If AI dominates invention, do incentives shift entirely toward machine builders?
  • Disclosure and credibility: Patent law requires enablement and full disclosure. If AI generates black-box innovations, can the human applicant sufficiently explain and support the invention?
  • Legal reform vs judicial interpretation: The Full Court ruling suggests that any change to allow AI inventors likely must come from Parliament, not courts.

These tensions mean that, despite excitement, reform will not be simple or immediate.

(Image: AI assistance and patent accountability)
(Image: AI assistance and patent accountability)

Where IP Guardian Comes In

At IP Guardian, we believe the AI inventorship debate is not just an academic point. It has real consequences for innovators building in biotech, clean energy, software, and more. Here’s how we help clients:

  1. Strategic drafting: We structure applications to emphasise human inventive contributions while acknowledging algorithmic assistance.
  2. Openness to novel claims: We monitor emerging AI and patent trends globally so clients are positioned to pivot quickly if law reform occurs.
  3. Global filing support: If clients want to file in jurisdictions more open to AI or experimental claims, we help bridge Australian protection with international strategies.
  4. Policy insight: We advise clients on government consultation processes, AI ethics frameworks, and how to frame their R&D pipeline for future legal shifts.

Because the debate continues to evolve, having a partner who understands both IP law and emerging technology is vital. IP Guardian is here for you. 

What happens next

While the current law in Australia bars AI from being inventor, change is not inconceivable. As AI systems become more autonomous, pressure will build for law reform, statutory updates, or regulatory adaptation. We expect to see:

  • Increased lobbying and consultation around adapting the Patents Act or regulations.
  • Test cases in niche fields (e.g. generative biology, autonomous design) pushing boundaries.
  • International divergence, with some jurisdictions leading reform and others resisting.
  • More complex claims involving AI assistance where human inventiveness may be subtle or hybrid.

For Australian innovators, the message is clear: protect what you can now, but keep an eye on how this frontier evolves.

The question “Can artificial intelligence be an inventor?” touches the heart of how we reward innovation. In Australia, the law currently says inventors must be human. The DABUS case reinforced that position, and the Full Court’s ruling has held firm. Yet the pace of AI development makes this a live topic, not a closed one.

At IP Guardian, we help inventors and organisations navigate these uncertainties. Whether your innovation includes AI, machine learning, or autonomous systems, we make sure your patent strategy is robust, forward-looking, and ready to adapt.

Curious how AI, patents, and your industry intersect? Reach out to our IP Guardian team for a consultation. We’re happy to explore possibilities with you.

Book a Consultation With a Trade Mark Attorney

ip guardians director - Can Artificial Intelligence Be an Inventor? The Ongoing Debate in Australia’s Patent System - IP Guardian Pty Ltd

Book a Free Consultation

We love our clients, so feel free to book a time to discuss your needs. The consultation is obligation and cost free!
Book nowContact Us

Related Posts

chevron-down