Supreme Court opens the door in the UK to patentability of AI inventions

Artificial neural networks

This dispute concerned a patent application for a system for providing media-file recommendations to end users based on an artificial neural network (ANN) that had been trained in a distinct manner. Initially, the hearing officer at the UKIPO had rejected the application, finding that the invention was a computer program "as such" and therefore was not patentable.

The High Court reversed this decision, holding that the ANN did not fall within the computer program patentability exclusion and it was not a computer program at all. The High Court's decision was then reversed by the Court of Appeal, which upheld the hearing officer's decision that the computer program exclusion applied and that inventions involving ANNs should be treated in the same way as any other computer-implemented invention.

The Aerotel approach

In coming to their decisions, the hearing officer, High Court and Court of Appeal each applied the Aerotel approach. This was a four-step test for determining whether the computer programs exclusion applied, which focused on identifying the actual contribution of the claim and assessing whether the contribution was technical in nature.

The hearing officer and the Court of Appeal found that Emotional Perception's invention was "no more than a computer program" that made no technical contribution and therefore was not patentable. Emotional Perception appealed to the Supreme Court.

Conformity with EPO case law?

The preliminary issue before the Supreme Court was whether the Aerotel approach should still be followed. The Supreme Court first re-iterated that the UK statute aims to bring UK patent law into conformity with the provisions of the EPC.

Although UK courts are not "strictly bound" by EPO Boards of Appeal decisions – even the Enlarged Board of Appeal (EBA) – they should "respect and follow" the EBA's decisions and any "uniform jurisprudence of the Boards of Appeal" unless the court is convinced that they are "wrong or beyond the ambit of reasonable difference of opinion". The Supreme Court noted that national courts of EPC contracting states should seek a uniform interpretation of the EPC and domestic legislation passed to implement it for "good policy reasons".

Aerotel had been decided by the English Court of Appeal when there was not a settled approach to the issue in EPO case law. Indeed, the Court of Appeal had noted three different approaches that had been adopted by the EPO.

Should Aerotel still be followed?

The computer program exclusion was not considered by the EBA until its decision in G1/19 in 2021. In that decision, the EBA made extensive criticism of Aerotel. In G1/19, the EBA adopted the "any hardware approach", which asks whether the claim involves the use of or is a to a piece of physical hardware, however mundane. If yes, the computer program exclusion will not apply. The Supreme Court noted that this "any hardware approach" is the first, necessary, stage in identifying whether there is an invention at all, before being able to move on to consider novelty and inventive step.

With this in mind, the Supreme Court adopted the EBA's approach in G1/19, bringing UK law in line with that of the EPO. It agreed with the EBA's criticism that the Aerotel approach was incompatible with the EPC because it was based on a misunderstanding of the meaning of the term "invention". In Aerotel, after construing the claim, the test goes straight to identifying the actual contribution of the claim and asking whether the contribution (rather than the invention) falls solely within the excluded subject matter. The Supreme Court noted that this framing raises considerations of novelty and inventive step, both of which are separate and independent requirements of patentability.

In endorsing the "any hardware approach", the Supreme Court rejected criticisms that the "any hardware" test was "window-dressing" because a competent patent drafter would be able to avoid the exclusion by including some piece of hardware in the claim. While the court accepted that the approach was "only a low hurdle", it held that this was justified by the inclusion of the "as such" wording in the exclusion.

Applying the 'any hardware' test

Is an ANN a computer program?

The Supreme Court first considered the definition of a computer and a computer program. The comptroller-general of the UKIPO had argued that a computer was merely a "machine that processes information", while Emotional Perception sought to narrow the definition to a digital computer containing a central processing unit.

While the court rejected the comptroller's definition as overly broad, it stated that there was no reason to confine the term "computer" to conventional digital computers. It reasoned that the concepts used in the EPC are "intended to accommodate technological change".

It also found that the description of a "program" must be correspondingly broadened to include a "set of instructions capable of being followed by a computer (of any kind)…to produce desired manipulations of data".

Taking into account these definitions, the Supreme Court agreed with the Court of Appeal that an ANN constitutes a set of instructions to manipulate data in a particular way and it is therefore a computer program. However, the Supreme Court disagreed with the Court of Appeal's distinction between hardware and software ANNs – whatever machine the ANN is implemented on will function as the computer.

Is the entire subject matter excluded?

While the Supreme Court found the ANN in this case to be a computer program, it was not a computer program "as such" because, applying the "any hardware approach", the ANN was implemented on "some form of computer hardware". The court noted that the claims referred to a database for storing data files, a communications network and a user device – all of which constituted hardware. This was sufficient to satisfy the low hurdle of the "any hardware approach".

Intermediate step

On its face, this decision is positive for those looking to patent AI-related inventions. It follows from the Supreme Court's reasoning that the patent should not be refused for excluded subject matter and the application must be re-examined by the UKIPO for novelty and inventive step.

However, there is still a measure of uncertainty because the Supreme Court held that there should be an "intermediate step" between the "any hardware approach" and the novelty and inventive step analyses. The intermediate step requires identifying the technical features of the invention (and excluding the invention's non-technical features) to then carry out novelty and inventive step assessments.

It is not yet clear how the intermediate step analysis will be applied in practice. The Supreme Court remitted the case back to the UKIPO, declining to perform the analysis itself because the test has not been applied in the UK before. This avoided creating binding precedent on an underdeveloped point of law.

That being said, the Supreme Court confirmed that the intermediate step requires a dissection of the subject matter of the invention into its component features and a review of the role of each feature to the technical character of the invention as a whole. It is not necessary to examine whether a feature is itself technical, but the "sole criterion" is whether the feature contributes to the invention's technical character.

It will now be for the UKIPO to carry out the intermediate step analysis and to determine whether Emotional Perception's invention meets the novelty and inventive step requirements.

Osborne Clarke comment

This decision represents a significant change to the way in which UK courts will approach the computer program exclusion. The Supreme Court's judgment means that there is now a low initial threshold for the patentability of AI-related (and other computer-implemented) inventions, which might make it easier to obtain patent protection. However, that will depend on how the UKIPO applies the intermediate step, as this could present a new hurdle to patentability.

Inventions will also still need to meet the separate novelty and inventive step requirements. The burden for applicants will now shift to demonstrating genuine technical contribution and inventive step. Businesses will need to adapt their patent drafting accordingly and must expect more rigorous examination of whether their AI inventions are genuinely novel and inventive.

All eyes will now be on the UKIPO – how will it apply the Supreme Court's decision and will it produce yet another version of its guidance on AI-related inventions? Uncertainty will persist until there is a decision from the UKIPO and further appeals are entirely possible given that this is new ground in the UK.

Beyond this case, it is not clear whether the Supreme Court's decision will spur further active alignment by UK courts with EPO case law on issues where there typically has been a difference in approach – for example, the issue of plausibility. However, there is less of a settled and aligned view at the EPO on plausibility than there was in this case and therefore there is greater scope to argue that there is a reasonable difference of opinion.