Can an AI invent? UK, European and Australian perspectives explained
Catherine Bonner, Terence Broderick
Patent offices around the world have been considering whether an AI — just like a human — can be a named inventor on a patent application. In this note, European Patent Attorney Dr Terence Broderick and trans-Tasman (Australia and New Zealand) Patent Attorney Arthur Roberts explore the contrasting views adopted by the UK, European and Australian patent offices and recommend a simple approach for those who feel they have an AI-developed invention.
The Turing prophecy
In 1951 — the year after publishing his seminal paper on artificial intelligence — legendary computer scientist Alan Turing noted “It seems probable that once the machine thinking method had started, it would not take long to outstrip our feeble powers”. Many interpret this as Turing predicting a machine takeover. Today, the cynics among us may argue that he was correct — we are already, for the most part, glued to various electronic devices. Recently, we may have taken another step towards Turing’s prophetic future, with patent offices around the world exploring whether an artificial intelligence can be regarded as an inventor — and therefore be awarded the same rights as a human inventor.
Does an inventor need to be human?
An international test case brought by Dr Stephen Thaler and his collaborators (creators of the AI machine DABUS) in a family of patents focused on whether an inventor needs to be human or not — and ultimately, in the latter instance, whether a patent application can proceed. This matter has been considered by multiple patent authorities including the UK Intellectual Property Office (UKIPO), the European Patent Office (EPO) and the Australian Patent Office.
The UK perspective
The UK is a little further ahead in its thinking than other jurisdictions on this problem, with the case having been heard by the UK Court of Appeal, following decisions by both the UKIPO and the UK High Court (on appeal from the UKIPO).
The Court of Appeal upheld the decision of the lower authorities by maintaining that an inventor must be a human being and that a patent application cannot proceed unless it has human inventors. It’s possible that a further appeal to the UK Supreme Court could be filed. If the UK Supreme Court did overturn the decision of the Court of Appeal, there could be substantial ramifications for the patent system, as it will acknowledge creative power of machines. Permission to appeal to the UK Supreme Court has now been sought.
Under UK law (and, specifically, section 7(3) of the Patents Act 1977), an inventor is the actual deviser of the invention — not merely an entity involved with the inventive process. While computers are often involved with research and development, they’re programmed to undertake tasks. Acknowledging that they themselves have devised anything gives them more than the capability to act on instruction — it gives them creative capability. Many in the software and computing community would argue that even the most powerful modern computers are still only acting on instruction.
The European perspective
Similarly, the EPO’s Legal Board of Appeal has also decided that an inventor must be a person with legal capacity (i.e., a human being) and that a patent application cannot succeed where the designated inventor is an AI, as it wouldn’t comply with the necessary EPO requirements. This decision upheld the initial view and decision from the Receiving Section of the EPO (which examines formality requirements for patent applications) which stated that an inventor must be a ‘natural person’.
We look forward to reading the decision and details when they are released. However, in its initial reasoning, the EPO outlined how inventors enjoy a series of benefits linked to inventorship that machines cannot share. These may include financial benefits (such as rewards in an employee inventorship program) and career development based on the kudos associated with inventorship.
Such ‘public policy’ perspectives indicate that the Legal Board of Appeal would prefer not to deviate too far from this reasoning as, firstly — the EPO typically doesn’t deviate very far from the ambit of patent law (perhaps correctly) and, secondly — providing a decision that deviates to the extent that AI inventorship is acceptable in some circumstances would open up other complex questions relating to whether an AI could be a patent applicant (as well as other queries).
The Australian perspective
Fascinatingly, back in July 2021, the Australian Federal Court decided that an AI system can be an inventor, placing it at odds with both the UK and European stance. Justice Beach found that there was “no specific provision in the [Patents] Act that expressly refutes the proposition that an artificial intelligence system can be an inventor” and further that there are no requirements for there to be a human inventor.
The Australian Patent Act provides no specific definition of an inventor, which means that the ordinary meaning of the word is used. “Inventor” is an agent noun just like “computer” or “dishwasher”. By looking at these and other agent nouns, we can see that an agent can be either a person or a thing — and therefore, so too can an inventor.
Further to the inventorship aspect, Justice Beach found no issue with Thaler owning the output of DABUS by virtue of him being the owner, programmer and operative of the AI system. Therefore, the invention was made for him “on established principles of property law”. An analogy was made to the ownership of “progeny of animals… produced by the labour and expense of the occupier of the land”.
Updates to the Australian Perspective (August 2022)
Recently, the Full Court of Australia overturned the Federal Court’s previous decision on AI inventorship, meaning that AI (or other non-human) inventors cannot be named as inventors of an Australian patent. This brings Australian practice in-line with the United Kingdom and European Union.
Our advice around AI inventors
This leaves South Africa as the only territory to have acknowledged the credibility of an AI inventor. Such inconsistency across patent jurisdictions is nothing new, but this latest instance could create problems at the earliest stages of the patent process — i.e., when inventors are named. For example, in some territories a patent application cannot proceed unless inventors are named. This prevents patent applications from moving forward before novelty and inventive step have even been considered.
It’s difficult to see the UK and Europe changing their perspective in the short term, given the strong legal reasoning on which their current position rests.
While we will continue to monitor this situation closely, if you feel that you may have an AI inventor, our recommendation is to ensure that you have at least one human inventor named on all your patent applications. This way, even if your AI inventorship is rejected by the respective patent offices, you will still be able to proceed with the patent application.
This article was originally published in February 2022 and updated in August 2022.