20 December 2023

UK Supreme Court Decision on AI Inventorship

The Supreme Court has today confirmed that artificial intelligence (AI) systems cannot currently be named as an inventor on a patent application in the UK. It has also confirmed that an owner of an AI system that creates a new invention is not entitled to a patent for that invention simply by virtue of owning the AI system.


This case concerns two UK patent applications filed in 2018 for inventions allegedly created by an AI system called “DABUS” (Device for the Autonomous Bootstrapping of Unified Sentience). The patent applications were filed by Dr Stephen Thaler (the creator and owner of DABUS), but DABUS was named as the inventor on each application.

Sections 7(3) and 13(2) of the Patents Act 1977 define the inventor of a UK patent application as the “actual deviser” of an invention and that the applicant must identify the “person” they believe to be the inventor. Section 7(2) also states that a patent may be granted primarily to the inventor or to any person entitled by agreement or rule of law, or their successor in title.

Applicants for patents at the UK Intellectual Property Office (UKIPO) are therefore required to file a statement of inventorship and, if the applicant is not the inventor, to indicate how they derived the right to be granted a patent.

In the case of Dr Thaler’s applications, he not only indicated that DABUS was the inventor but also maintained that he himself was not an inventor. He asserted that he was entitled to grant of the patents by virtue of owning DABUS.

At a hearing in November 2019, Dr Thaler’s arguments were rejected by the UK Intellectual Property Office on two grounds:

(a) DABUS is non-human so is not a “person” as envisaged by section 7 or section 13 of the Patents Act, and therefore cannot be an inventor, and

(b) Dr Thaler is not entitled to apply for these patents simply by virtue of owning DABUS, since DABUS – not being a “person” – had neither any rights that could be transferred, nor any power to transfer anything that it might have owned.

Accordingly, the applications were held to be withdrawn upon expiry of the period for providing the statement of inventorship.

Dr Thaler appealed this refusal, but his appeal was dismissed in the High Court and subsequently the Court of Appeal in September 2021 (on a 2:1 majority). This was then appealed further to the Supreme Court, who have now also dismissed the appeal.

The Judgment

The Supreme Court considered this appeal to be concerned with the relatively narrow question of how the relevant provisions of the Patents Act 1977 should be interpreted, rather than a broader policy consideration of whether the meaning of the term “inventor” should be expanded to allow AI systems to be named as inventors.

On the first issue of whether DABUS could be named as an inventor, the Supreme Court unanimously found that an inventor must be a natural person. Lord Kitchin noted that, at the time of drawing up the Patents Act, Parliament did not contemplate the possibility that an AI system acting on its own could devise an invention, so the laws are not intended to cover this possibility.

The second issue, namely whether Dr Thaler was entitled to the grant of a patent for the inventions created by DABUS, has also been answered with a “no”. Lord Kitchin explained that since DABUS cannot be and has never been an inventor, and that Dr Thaler has himself admitted that he is not an inventor, there is no link between Dr Thaler and an inventor (i.e. no link of the type provided for in the Patents Act, such as an employer-employee relationship or other succession in title). As the Patents Act provides a “complete code” for the purpose of determining the right to apply for and obtain a patent, which starts with the inventor, then the absence of an inventor must cause the application to fail.

Although Dr Thaler argued that as the owner of DABUS he was entitled to the grant of patents for inventions created by it under the doctrine of accession (in the same way that a farmer owns a calf born to his cow), this argument was also dismissed. The Court held that on the basis that an invention is not tangible property, there is no basis for the application of the principle of accession.

Finally, the third issue considered was whether the UKIPO was correct to treat Dr Thaler’s applications as withdrawn. The Supreme Court agreed with the findings of the lower courts that Dr Thaler did not meet the requirements of Section 13(2) of the Patents Act, so his applications were rightfully deemed withdrawn.


This judgment is consistent with the current laws on inventorship and entitlement, although as flagged by Lord Kitchin, this is partly due to the fact that they were drawn up some time ago before issues of AI inventorship would have been envisaged. However, with the increasingly rapid development and adoption of generative AI, these issues are unlikely to go away any time soon.

It is worth noting that the Supreme Court emphasised that neither the UKIPO nor the courts investigated Dr Thaler’s assertions of fact, namely that the inventions were generated autonomously by DABUS and that Dr Thaler was not an inventor. Indeed, it is not the function of the UKIPO to investigate the factual correctness of such statements. As recognised by the Supreme Court, the outcome may well have been different if Dr Thaler had asserted that he was an inventor who had merely used DABUS “as a highly sophisticated tool”.

Although corresponding cases in Australia and South Africa found that an AI machine could be named as an inventor, for the time being applicants in the UK will therefore need to identify the human inventors who have devised the invention. In the context of AI-generated or AI-assisted inventions, this still leaves several grey areas in determining who should be named as an inventor – for example, whether it should be the creator, owner or user of the AI system, or even someone who has identified the inventive concept of an invention created by an AI system.

We look forward to seeing how patent law in the UK and overseas continues to respond to challenges posed by AI.