In a landmark decision on 11 February 2026, the UK Supreme Court has fundamentally reshaped the UK's approach to the patentability of computer-implemented inventions. In positive news for patentees, the Judgement overrules almost 20 years of patent practice and confirms that qualification as 'an invention' in the UK is possible for claimed subject matter that requires any hardware at all. This aspect of the decision will lower the hurdle for obtaining patent protection in the UK for AI and computer-implemented inventions, so long as the patent involves the use of any piece of physical hardware, however mundane.

For the appellant, Emotional Perception AI (EPAI), an interpretation of their claimed artificial neural network (ANN) as a program for a computer as such (as proposed by the UK Intellectual Property Office and endorsed by the Court of Appeal) is rejected by the Supreme Court, as the ANN "can only be implemented on some form of computer hardware".

This view is consistent with the approach of the European Patent Office (EPO), and rejects the long-established Aerotel methodology previously used by the UKIPO. However, the assessment of novelty and inventive step will require consideration of the presence of technical character, and the Judgement does not offer guidance on how the UK should approach this issue. The substantial body of EPO case law may be persuasive, and harmonisation between the UK and EPO approaches is to be welcomed.

How did we get here?

EPAI applied for a patent for an ANN-based system that uses machine learning to provide media file recommendations based on emotional or semantic similarity. The application had a tortuous journey through the courts:

• UKIPO Hearing Officer (2022): Rejected the application, holding that the claimed invention was excluded from patentability as a "program for a computer ... as such" under section 1(2)(c) of the Patents Act 1977.
• High Court (November 2023): Sir Anthony Mann allowed EPAI's appeal, holding that an ANN is not a program for a computer at all, and therefore the exclusion was not engaged. This High Court decision was discussed in our 2023 blog post here.
• Court of Appeal (July 2024): Birss LJ, with Arnold and Nicola Davies LJJ agreeing, allowed the Comptroller's appeal and reinstated the Hearing Officer's decision, ruling that ANNs are computer programs and that the invention lacked a technical contribution. We discussed the Court of Appeal decision on our blog here.

Three key issues before the Supreme Court

EPAI were granted leave to appeal the decision of the Court of Appeal to the Supreme Court, with our earlier blog post summarising the arguments made by the parties at the hearing. Three key issues arose in consideration of EPAI's appeal:

1. Should the Aerotel guidance no longer be followed?
2. Is an ANN a "program for a computer"?
3. Is the entire subject matter of the claims excluded?

Issue 1: Should the Aerotel guidance no longer be followed?

The principles set out in Aerotel Ltd v Telco Holdings Ltd [2006] EWCA Civ 1371 ("Aerotel") have governed UK patent practice for almost 20 years on the issue of exclusions from patentability. Section 1 of the Patents Act 1977 sets out a range of exclusions from patentability (including scientific theories, mathematical methods, business methods or a program for a computer), but "only to the extent that a patent or application for a patent relates to that thing as such". Under the Aerotel approach, it is determined whether the actual contribution of an invention to the state of the art falls solely within excluded subject matter, and leads to rejections of inventions in which the actual contribution is considered to be a program for a computer as such.

The Court noted that Aerotel was decided at a time when the approach to applying Article 52 of the European Patent Convention (EPC) had not been crystallised by the EPO. By way of reminder, the European Patent Convention is a multinational treaty to which the UK is a member, which creates the EPO, a Europe-wide legal system under which European patents are granted and which partially harmonises the domestic patent law of its member states. The Supreme Court referred to the observation of Jacob LJ in Aerotel that "if and when an Enlarged Board did rule on the question, the approach in the UK might have to be reconsidered", and looked to the reasoning of the EPO's Enlarged Board of Appeal in its decision of G1/19.

In doing so, the Court endorsed the EPO's "any hardware" approach, under which the subject matter of a claim will not be excluded from patentability under Article 52(2)(c) EPC if it embodies or involves the use of any piece of physical hardware, however mundane. In G1/19, the Enlarged Board criticised the Aerotel approach as incompatible with the EPC and based upon a misunderstanding of the word "invention" as used in Article 52 EPC.

The Supreme Court held that the Aerotel approach should be abandoned. However, it rejected the Comptroller's concern that adopting the EPO's approach would require abandoning the established Pozzoli test for inventive step in favour of the EPO's "problem and solution" method. The Supreme Court confirmed that Pozzoli remains a legitimate approach to assessing inventive step.

Issue 2: Is an ANN (or does it contain) a "program for a computer"?

EPAI's primary argument, and the basis on which they obtained favourable judgement at the High Court, was that an ANN is not a program for a computer, and the computer program exclusion does not apply to their invention.

The Supreme Court rejected this argument, and disagreed with EPAI's submission that the term "computer" should be confined to conventional digital computers with a CPU. The Court argued that whatever the specific form of the physical machine on which an ANN is implemented, the ANN is a set of instructions to manipulate data in a particular way to produce a desired result. It is therefore a "program for a computer" within the meaning of Article 52(2)(c) of the EPC. Put simply, the fact that an ANN can be implemented in a variety of different types of hardware, demonstrates that an ANN cannot itself be regarded as a type of hardware.

However, the fact that an ANN is a program for a computer does not, under the reasoning of the Supreme Court, lead to the conclusion that the claimed ANN is to be excluded as a program for a computer as such.

Issue 3: Is the entire subject matter of the claims excluded?

The adoption of the "any hardware" approach leads to the conclusion that the claims are to an "invention" and would not have been excluded by Article 52(2)(c) EPC. The ANN can only be implemented on some form of computer hardware. The claims also refer to a database, a communications network, and a user device - all of which require or constitute hardware. This is sufficient to show that the subject matter has technical character and is not a claim to a computer program "as such".

The "intermediate step" - what happens next?

Although the patent application clears the "any hardware" hurdle, the Supreme Court declined to decide whether the invention ultimately qualifies for a patent. Following the guidance in G1/19, there is an "intermediate step" that must be undertaken between determining that a claim qualifies as an "invention" and assessing novelty and inventive step.

This intermediate step requires filtering out features that do not contribute to the technical character of the invention as a whole, so that only features which do contribute are considered at the inventive step stage. The Court explained that this is fundamentally different from step 4 of the Aerotel approach, which asked whether the contribution of the invention as a whole was technical in nature.

However, since the case had been argued entirely on the basis of Aerotel, the Court received no detailed submissions on how the intermediate step should be applied. It therefore referred the case back to the Hearing Officer to apply the intermediate step and consider the remaining requirements of novelty and inventive step.

Analysis

The Supreme Court's decision, whilst welcome in principle, leaves significant practical questions unanswered for patent applicants and practitioners:

The intermediate step remains undefined: The Supreme Court expressly declined to provide complete guidance on how the "intermediate step" should be implemented in the UK. The Court acknowledged that this appeal raised issues of "fundamental importance" for UK patent law, but stated it would be "unwise" to create binding precedent on the intermediate step without the benefit of in-depth legal argument and the views of the UKIPO and specialist judges.

Feature-by-feature analysis is new territory: The intermediate step requires a feature-by-feature dissection of the claimed invention to identify which features contribute to its technical character - and which should be filtered out before assessing inventive step. This stands in "sharp contrast" to step 4 of the Aerotel approach, which required a holistic assessment of whether the contribution of the invention as a whole was technical in nature. The UKIPO has no experience applying this new methodology. It is difficult to predict with certainty how an examiner will resolve a claim into technical and non-technical features, and assess which features contribute to the technical character of the invention when viewed as a whole. The Supreme Court confirmed that the UKIPO need not adopt the EPO's "problem and solution" approach to inventive step, and suggested that identifying the "inventive concept" per Pozzoli might be an appropriate method in some cases, but this was expressly not a ruling on the point.

Reliance on EPO case law: Given the absence of UK guidance, the substantial body of EPO case law applying the Comvik approach is likely to be highly persuasive - but it is not binding, and the UK courts may develop their own interpretation of the intermediate step over time. Applicants may face a period of uncertainty whilst the UKIPO and courts work through these issues.

EPAI's application remains undecided: For EPAI itself, the outcome remains uncertain. The case returns to the Hearing Officer to apply the intermediate step and consider novelty and inventive step. Whether features such as the "semantic similarity" of the recommended files - which the Court of Appeal characterised as "subjective and cognitive" rather than technical-will survive the intermediate step remains to be seen.

Summary

The Supreme Court's decision marks a watershed moment for UK patent law relating to AI and computer-implemented inventions. Key points include:

1. Abandonment of Aerotel: The four-step Aerotel approach is no longer good law. UK courts must now apply the "any hardware" approach endorsed by the EPO in G1/19.
2. ANNs are computer programs: An ANN is a program for a computer, regardless of whether it is implemented in hardware or software. The weights, biases, network topology, and activation functions all operate in combination to instruct the machine to process data in a particular way.
3. Lower first hurdle, but patentability not assured: The "any hardware" test imposes a very low hurdle for qualifying as an "invention". However, this does not guarantee patentability - the intermediate step and full assessment of novelty and inventive step remain to be undertaken.
4. Pozzoli survives: The established UK approach to inventive step, as set out in Pozzoli and approved in Actavis v ICOS, remains valid. Adoption of the EPO's interpretation of Article 52 EPC does not require wholesale adoption of the "problem and solution" method.
5. Guidance for the future: Whilst the Supreme Court has provided significant clarity on the correct interpretation of Article 52, detailed guidance on implementing the intermediate step in the UK will need to be developed by the UKIPO and the lower courts in future cases.

For EPAI, the battle is not yet won - its application returns to the Hearing Officer for further consideration. For the wider patent community, this decision signals a fundamental realignment of UK practice with EPO jurisprudence on computer-implemented inventions, whilst preserving the UK's established methodology for assessing inventive step.