The English High Court has upheld a UK IPO ruling that an AI machine cannot be an inventor under UK patent law. As AI continues to change the world around us, will this position remain? 

Earlier this month the UK IPO put out a call for views to help it understand the implications AI might have for IP policy. In a bid to ensure the UK remains at the forefront of the AI and data revolution, the IPO set out a list of questions about the intersection of AI and IP and has invited input and insight into how best to make sure the UK’s IP environment is adapted to accommodate and encourage growth in transformational new technology sectors.

With the IPO’s AI public outreach still hot off the press, the English High Court this week published its decision relating to one of the key questions in the field of patent law circulated for debate by the IPO – namely, whether AI can be a patent inventor. Marcus Smith J ruled that an AI machine called DABUS  “is not, and cannot be, an inventor within the meaning of the 1977 [Patents] Act, simply because DABUS is not a person”.

The judgment 

The case was an appeal by Dr Stephen Thaler (the creator of the DABUS machine) whose two patent applications, which designated DABUS as the inventor, had been rejected by the UK IPO. Parallel applications to the EPO (see our earlier blog on that decision) and USPTO had also similarly been refused. Of fundamental importance was that DABUS was a machine, whereas the legal frameworks applied by the UK IPO, EPO and USPTO require the inventor to be a natural person, or human.

Dr Thaler appealed the IPO’s decision. In his submission’s Dr Thaler maintained his position that DABUS was the inventor and denied that status for himself. The resulting key question was a simple one: could an AI machine be regarded as an inventor under the Patents Act 1977 (the Act).

As a starting point, while Smith J acknowledged that the court “cannot itself legislate, no matter how great the policy need”,  it may well be that courts without re-writing law can interpret the law to accommodate new technologies: “the common law or a scheme laid down in statute does — when appropriately construed or understood — cater for future developments, including developments that were — until they surfaced in litigation — unforeseen”.

The judge went on to find that the legislation was quite clear that “whatever the meaning of the term "inventor" [in the Act] — a patent can only be granted to a person” on the basis that: (i) only a person can hold property and an invention, and an application for the grant of a patent and the patent itself are all property rights; and (ii) the language of the Act makes clear that the holder of a patent must be a person.

The Act says that the ‘inventor’ is ‘the actual deviser of the invention’. The Judge held the ‘inventor’ is the “person who is the actual deviser of the invention” and that the natural reading of the Act is “that the inventor is a person and the invention a thing”.

Consequently, Smith J ruled that even though DABUS could well have “invented the inventions the subject of the Applications. Nevertheless, I conclude that DABUS is not, and cannot be, an inventor within the meaning of the 1977 Act, simply because DABUS is not a person”.

Finally, the Judge said that even if he had found that DABUS was capable of being an ‘inventor’, seeing as it wasn’t a person, it would be “incapable of conveying any property to Dr Thaler… because DABUS is a thing, it cannot even hold property, let alone transfer it”.

Open questions 

The judgment emphatically concludes that under the current statutory framework, AI cannot be an inventor. Smith J was keen to point out (he did so on a number of occasions in the judgment) that that he had not considered, and that no deductions should be made from his ruling, as to the question of whether the owner of an AI machine that invents can themselves be the inventor. Indeed, the judge seemed to indicate a court would be open to such an argument “it would be far easier to contend that Dr Thaler was entitled to the grant of the patent… on the ground that he (Dr Thaler) owned the machine that did the inventing”. In other words, the machine did the thinking for both of them.

Dr Thaler expressly refused to make such a claim, in the belief that this is both fundamentally wrong (i) legally – the true inventor is the AI machine; and (ii) morally – it misleads the public and by allowing individuals to take credit for the work of machines. Dr Thaler claimed that other human inventors’ work and accomplishments would be diminished by altering the meaning of inventorship. Dr Thaler appears to be good at being noble and both are reasonable arguments. In the IPO’s initial DABUS decision, it concluded that "AI machines are likely to become more prevalent in the future and there is a legitimate question as to how or whether the patent system should handle such inventions" because "times have changed and technology has moved on". Similarly, the IPO has called for a wider debate on any changes to the law. Although the question may be better for the legislator and not the judiciary.

This leads us to the IPO’s recent call for views on AI and IP and their list of questions. Notably, the DABUS hearing and questions it raises was specifically referenced by the IPO as being a paradigm example of the open issues. 

Discussions on these questions will be hotly debated. Just taking this issue of whether AI should be considered an inventor: on the one hand, some argue (as outlined by the IPO) that AI systems are no different to other tools that are used by people to invent. Questions have been asked about why AI algorithms are different from other platform technologies such as combinatorial chemistry which is used to produce novel drug compounds.

On the other hand, not affording patent protection to AI machines could discourage future development of AI innovation, or perhaps more likely lead to the increase in the use of trade secrets and a decrease of flow of information and technological advancement (which is supposed to be encouraged by the patent system). Excluding AI machines from being inventors might simply incentivize concealment of the involvement of AI to circumvent the rules. That would seem an undesirable outcome, especially given the recent push across different fields to demystify AI, as exemplified by the UK Information Commissioner's Office guidance to help businesses and organisations explain decisions made by AI systems (for further details on the ICO’s guidance see our blog here). 

Should it be decided that the AI can indeed be an inventor, a web of related questions ensues. If AI is the inventor, who or what should be entitled to own the patent? How can the AI machine transfer the patent if it is not a person?  These are difficult questions, even before one considers issues of patentability regarding AI-generated inventions. For instance, how can traditional tests of obviousness be applied to an AI-generated invention? (for further analysis on those questions, see our previous blog here).

There is a long list of open questions – certainly more than a hill of beans – so it is no wonder the IPO has opened the floor to views from industry and interested professionals. The ensuing discussions, conclusions drawn and ultimately the policy/statutory framework set by the IPO could go a long way to helping the government achieve its aim of making sure that the UK has the best environment for developing and using AI. Here’s looking at you, DABUS - watch this space.

For more information on the legal issues surrounding AI, click here.