AI Inventors and beyond – the Court of Appeal has spoken

AI Inventors and beyond – the Court of Appeal has spoken

The eagerly awaited judgement of the English Court of Appeal has come out this week. While the three judges agreed that an inventor has to be a person and the deemed withdrawal of Dr Thaler’s DABUS application has been upheld, one of the judges dissented from this conclusion, finding that the Patents Act does not require an inventor to grant a patent. While the actual outcome may not be that surprising, the split judgement on such a fundamental question of patent law whether patents can be granted for inventions without inventors may leave the door open for Dr Thaler to appeal. We will see if leave to appeal is thought and granted…

In the meantime, the judgement makes for a good read, see here.

If you don’t want to read the judgement in its full glory and/or are interested in my thoughts, here are some notes:

The judges agreed that an inventor, as the actual deviser of the invention, is a person. This is clear from the meaning of “deviser”, the legislative history of the sections in the patents act, and its internal consistency.

The judges agreed that sections 7 of the Patents Act, which deals with inventors and the right to obtain a patent, and section 13 of the Patents Act, which deals with the formalities to be complied with in respect of this, do not require the Patent Office to examine the merits of any statements in relation to who the inventor is or how rights have been derived by the applicant. What is more, the legislative history of the Patents Act and its provisions regarding entitlement challenges make it clear that the Patent Office does not have jurisdiction for such an investigation. This is where the agreement ended, though.

The majority of the judges (LJ Laing and LJ Arnold) held that the Patents Act requires that a person believed to be the inventor is identified, as well as a statement as to how the rights were derived by an applicant other than that person, failing which the application is deemed withdrawn. Dr Thaler has not identified such a person by identifying DABUS, the machine. As acknowledged by Dr Thaler, DABUS is not a person. There is also no mechanism known to English law by which any property in an invention created by a machine could end up with Dr Thaler simply by ownership of DABUS. This was not an examination of the merits of Dr Thaler’s statement but rather based on taking these statements at face value. Therefore, the Patent Office was right to deem the application withdrawn. This is also consistent with section 7 of the Patents Act being a self-contained set of rules as to how the right to a patent can arise, as held in previous case law. The appeal was dismissed.

A key quote from LJ Arnold for the majority opinion:

“Dr Thaler did not identify ‘the person or persons whom he believes to be the inventor or inventors’ as required by section 13(2)(a). On the contrary, he deliberately identified a non-person as the inventor. The fact that he may genuinely have believed that DABUS was the inventor is neither here nor there. The answer he gave to the question was a legal impossibility… Nor did Dr Thaler identify ‘the derivation of his right … to be granted the patent’. He simply asserted that it was sufficient that he owned DABUS. As a matter of law, that is incorrect.”

The dissenting judgement also makes for interesting reading. The dissenting judge (LJ Birss) on the contrary held that section 13 requires only that the applicant states their belief as to who the person that is the inventor is. By mentioning DABUS, Dr Thaler has stated his belief as to the person that is the inventor (there is none) and he has indicated how he derived the rights. Dr Thaler has therefore discharged his obligation under section 13 and it is not for the Patent Office to investigate the merits of the statements made. The application should therefore not be deemed withdrawn. The absence of an inventor is not an impediment to the grant of a patent and the statute provides that whether the applicant indeed has the right to be granted a patent can only be challenged by those alleging to be entitled in place of the applicant, not by the Patent Office or the courts of their own motion.

While the dissenting judge has certainly provided an interesting perspective on the provisions of the Patents Act, I am personally with LJs Laing and Arnold who held that reading the Patents Act it cannot be thought that inventions created by machines were contemplated by parliament and that therefore sections 7 and 13 have to be given its plain meaning. It follows that section 7 only can give rise to any rights if there a person who devised the invention, even if the rights maybe with another person due to a prior agreement or due to employment of the inventor. Similarly, the plain meaning of section 13 is that a person that is believed to be an inventor has to be identified and that this requirement cannot be discharged by identifying a non-person.

As stated in the majority judgement, if inventions made by machines rather than persons are to be protected by patent law, this is not something previously contemplated by Parliament and is for our legislature to consider and legislate for if it decides to do so. The Patent Office is currently consulting on this topic and it remains to be seen if it makes it onto Parliament’s agenda.

Until and unless the matter gets picked up by Parliament, this will be it for DABUS in the UK, unless of course Dr Thaler is granted leave to appeal the matter to the Supreme Court. Lastly, I think the opening statement of the dissenting judgement points to a wider issue:

“ At first sight […] the case appears to be about artificial intelligence and whether AI-based machines can make patentable inventions. In fact, this case primarily relates to the correct way to process patent applications through the Patent Office [...]. It is an object lesson in the risks of advocacy being distracted by glamour.”

Indeed, the broader question of whether a machine has in fact the capability to devise an invention (and what this means) has not been considered by any forum I am aware of in this or another country (see here, here, here, and here). Judging by the lack of engagement from the academic and industrial AI community with DABUS, and vice versa, maybe it is premature to ask whether inventions devised independently by AI require patent protection or not.

Read previous article | Read next article

Follow Alexander Korenberg’s AI musings

AI musings