Legal Outlook Podcast 3

Legal Outlook podcast 3: transcript

The PDF server is offline. Please try after sometime.
  • My Documents


    Material personally selected by your relationship manager for your interest.

  • My Bookmarks


    Access all of the content that you have previously selected to bookmark.

  • Get started


    ​Scroll through these slides to access the personalised features of your Dashboard.

  • My Suggested Reading


    A virtual library of regularly posted insights and legal updates based on your selected preferences.

Get Started


Eoin Martyn:
Hello, and welcome to episode three of Ashurst Legal Outlook. My name is Eoin Martyn, Senior Associate, IP Media at Ashurst. This episode is the third in a series dedicated to artificial intelligence. In this series, you will hear global perspectives along with local Australian examples, as we explore and unlock the mysteries of AI and its legal and business implications. In today's episode, I'm joined by Nina Fitzgerald, IP Media Partner at Ashurst to discuss all things AI and patents. Nina, and I will discuss the value of patent protection and patentability of AI in Australia, what the future of AI looks like under Australian patent law. You are listening to Ashurst Legal Outlook. Hello, Nina. So far in this series, we have introduced AI to our listeners, and discussed the ethics around AI. In this episode, we're going to look more specifically at AI and patents. Can you firstly, explain to our listeners the value of obtaining patent protection and the object of the patent system?

Nina Fitzgerald:
Thanks Eoin. Our law typically favors competition, and this is valued because it encourages business to flourish and reduces costs to consumers. However, governments have also recognized that there are significant costs to companies and individuals in inventing new products and processes. So in recognition of this cost and to encourage innovation, government have passed legislation to enable patents to be granted. Patents are an exception to the prohibition against monopolies, and grants an inventor a 20 year monopoly over inventive, new and useful products and processes. In exchange for this monopoly, patentees must disclose their inventions to the public. This enables other businesses to use that information after the patent period has expired.

Eoin Martyn:
Okay, in terms of patentability of AI, can AI and machine learning systems be patented in Australia?

Nina Fitzgerald:
This is somewhat of a fraught topic and potentially worthy of its own podcast, but in order to summarize it very quickly, to be patentable subject matter an invention must constitute a manner of manufacturer. This phrase is quite archaic, but has been interpreted by the High Court to mean that the invention must comprise an artificial state of affairs of economic utility, another very broad and potentially unhelpful saying. As a result, the courts have been interpreting what a manufacturer is over time by applying this test laid down by the High Court. This has led to laws of nature, mathematical equations, and business methods being found not to constitute a manner of manufacture and not to satisfy the test, thereby not being patentable subject matter. So this is where computer implemented inventions come in. Obviously AI systems operate through computers. However, the courts have been increasingly critical of computer implemented inventions, as they've consistently held that the patents for those inventions are actually for business methods, which are implemented in the computer and therefore fail a manner of manufacturer test

Eoin Martyn:
But Nina, does this mean that you can't get a patent for AI systems?

Nina Fitzgerald:
Absolutely not. This doesn't mean patents for AI systems are not being granted, rather the number of patent applications for computer implemented inventions are increasing rapidly. However, the court requires that a stringent set of criteria are reached before such patents are granted. In particular, the invention must satisfy a technical problem with a technical solution, so there can't just be a business problem which is implemented in the computer. Rather, the computer must be integral to the invention, and the computer can't simply employ steps that are already within its traditional functioning. You have to be making the computer do something new in order for a patent to be granted for that invention. So hopefully that's a relatively short summary of what is a very complex area of law.

Eoin Martyn:
Indeed, and it's certainly something that is continuing to evolve. Nina, as you and I are both aware, much discussion recently has focused on the ownership and entitlement to inventions which are created by AI. I was hoping you might update the listeners regarding the recent DABUS patent application.

Nina Fitzgerald:
Absolutely DABUS is an AI system. It was created by Stephen Thaler, an inventor in the UK, and he created the system, not for any particular purpose or to solve any particular problem, and as a result, it was trained in very general information from various domains, and as a result of this training, it conceived two quite different inventions. The first was for a special food container, and the second was a flashing light for use in emergencies. So as you can see two very different domains and industries for application. So Stephen Thaler identified these two inventions that were created by DABUS, and filed patent applications at multiple patent offices around the world, including the US PTO, the European Patent Office, the UK Intellectual Property Office and the Australian Patent Office. In each of these patent applications, the inventor was an AI machine called DABUS and Stephen Thaler himself was named as the applicant.

Eoin Martyn:
I will be interested to know, Nina, did the Australian Patent Office differ in their approach to the other offices, including the US PTO, EPO, and the UK IPO?

Nina Fitzgerald:
Interestingly, all patent offices have refused the patent application to date. The primary reason for this outcome in each jurisdiction was the fact that the inventor named in the patent application was DABUS, and DABUS was a machine. Whereas the legal framework in each of those jurisdictions require the inventor to be a natural person or a human. And if we look particularly in Australia, the Australian Patent Act requires the invention to be granted to, and the legislation states this, "a person who is the inventor, has been assigned the invention, has derived title to the invention, or is the legal representative of a deceased person falling into one of those previous categories." In circumstances where AI is not a person and cannot assign their rights to a person, AI cannot currently be listed as an inventor.

Eoin Martyn:
Interesting. So our listeners would be curious to know, looking from here to the future, what do you think needs to be done in Australia regarding AI generation work and inventions, and what can be done?

Nina Fitzgerald:
Well this raises some really interesting questions regarding whether patent laws should be revised to enable inventions created by AI systems to be patented. Going back to where we started at the beginning of this episode, we outlined the purpose of the patent system, being a fundamental means of encouraging investment in innovation and the sharing of knowledge to enable competition after the monopoly period is ended. Some have argued that the failure to reward inventions created by AI with a monopoly, could discourage owners from pursuing further development or result in owners protecting AI created inventions as trade secrets, without disclosing them to the public. In these circumstances, and if it is found that this is accurate, IE that people are not investing in AI or more likely owners, don't disclose AI created inventions, there would be a strong argument that there should be a reward for investment and disclosure of AI created inventions. The question is whether the patent system is the right means to protect AI created inventions, whether we need a new right, or whether we can fit AI created inventions into the existing patent scheme, and that may raise some very interesting questions around how we establish novelty, inventive step, who the appropriate person skilled in the art would be, if the invention has been created by an AI system. So this issue is currently being considered in many countries around the world, and we can expect it to continue to be a hot topic going forward.

Eoin Martyn:
Nina, this has been very interesting. I think the key insight for our listeners based on our discussion, is that patents can be incredibly valuable in that they are an enforceable and government-sanctioned right to commercially exploit an invention in the marketplace to the exclusion of all others, and the monopoly can last for up to 20 years. However, as you've pointed out, patents are not granted automatically. Patents for computer implemented, inventions need to be considered very carefully as they may be rejected. It may be more appropriate to protect such inventions using other means such as trade secrets, which will be covered in a future episode. Finally, I think companies and organizations need to be careful about claiming that an AI system could be listed as an inventor on a patent application, as such statements may risk subsequent patent applications being granted, as AI systems are not recognized as an inventor. So on that note, I'd like to thank Nina for joining me on the podcast. It's been an incredibly insightful episode, and I think we all have quite a lot to think about when it comes to AI on patents.

Nina Fitzgerald:
Thank you so much, Eoin. It's been my pleasure.

Eoin Martyn:
Thank you for listening. To hear more Ashurst podcasts, including our dedicated channel on all things AI, please visit To ensure you don't miss future episodes, subscribe now on Apple Podcasts, Spotify, or your favorite podcast platform. While you're there, please feel free to keep the conversation going and leave us a rating or a review. Thanks again for listening, and goodbye for now.

All Episodes

Key Contacts

The information provided is not intended to be a comprehensive review of all developments in the law and practice, or to cover all aspects of those referred to. Listeners should take legal advice before applying it to specific issues or transactions.


        Forgot Password - Ashurst Account

        If you have forgotten your password, you can request a new one here.


        Forgot password? Please contact your relationship manager to find out more about our client portal.
        Ashurst Loader