Hepworth Browne

Artificial Neural Networks and UK Patent Law: What Every AI Developer Needs to Know

Can artificial neural networks be patented in the UK? The Emotional Perception Supreme Court ruling changed the legal position for ANNs significantly. Here is what AI developers and businesses need to know.

Published14 August 2026
TopicAI
FieldArtificial intelligence, IPC G06N
Reading time12 minutes

Artificial neural networks are at the core of modern artificial intelligence. From large language models to image classifiers, from recommendation engines to medical diagnostics, the neural network architecture is the computational engine driving commercial AI deployment. The question of whether those architectures, and the inventions they enable, are patentable in the UK has been answered by the Supreme Court in Emotional Perception AI Limited v Comptroller General of Patents [2026] UKSC 3. Issues do, however, remain.

The ruling is good news for AI developers, and for industries that develop computer implemented inventions. The decision establishes that ANN based inventions are not automatically excluded from patent protection in the UK, that the correct analysis involves a properly structured multi stage assessment, and that the threshold for clearing the statutory exclusion has been significantly lowered to the presence of any hardware. But the ruling also clarifies that an ANN is itself a program for a computer within the meaning of UK and EPC patent law, a finding that has implications for claim strategy, specification drafting and prosecution approach. That finding is not ideal and is technically questionable, so one cannot rule out it being challenged or corrected, possibly by a very specific piece of legislation.

Is an artificial neural network a computer program under UK patent law?

Yes. The Supreme Court held that an ANN, whether implemented in dedicated hardware or emulated in software on a conventional computer, is a program for a computer within the meaning of Art.52(2)(c) EPC and section 1(2)(c) of the Patents Act 1977.

The court rejected the argument that a hardware ANN is a physical machine rather than a computer program, and that an emulated ANN, being functionally equivalent, should be treated in the same way. This is questionable logic and potentially legal abstraction. The court held that an ANN is properly described as an abstract entity or computational model, not a physical object. It can be implemented on various types of hardware, including neuromorphic chips, field programmable gate arrays and conventional digital computers, but it is not itself a type of hardware. It is the abstract arrangement of neurons, weights, biases and activation functions that constitutes the ANN, and that arrangement is in essence a set of instructions to manipulate data in a particular way.

The court reasoned that, at the end of training, the weighting factors on aggregation constitute a set of instructions for execution by the implementing hardware, wherein the individual neurons process data to produce a particular output. That instructions are created through a machine learning process rather than written by a human programmer does not, on the court’s analysis, affect their character as a computer program. The court noted that it is commonplace for computer programs to generate other programs. That, unfortunately, appears to represent a disingenuous mischaracterisation of human ingenuity. Neural networks are trained against learning objectives that sit at a conceptually higher level of generalisation, which contrasts with the ability of existing large language models to generate a program by selecting and knitting together core program modules from a library.

Does classifying ANNs as computer programs mean they cannot be patented?

No, and this is a critical point. That an ANN is a program for a computer merely engages the computer program as such exclusion of Art.52(2)(c), as tempered by Art.52(3) EPC. This first hurdle requires further qualification.

Under the any hardware approach now adopted, any claim that involves hardware, which any ANN implementation necessarily does, is not a claim to a computer program as such. The presence of hardware, however conventional, anywhere within the claim is sufficient to take the claim outside the statutory exclusion, so a computer implemented method or a processor based ANN achieves that objective.

ANN based claims are therefore not excluded at the Art.52 EPC or section 1(2) stage. What then matters is whether the invention satisfies novelty and inventive step, which requires an understanding of what amounts to the invention. An intermediate step between the exclusion analysis and inventive step investigates the features of a claim that interact with one another to contribute to its technical character, or more appropriately the technical effect achieved by the invention. This is a forward thinking analysis.

The intermediate step is therefore capable of differentiating true technical invention from sound engineering, the latter manifest for example in the straightforward automation of a business process. In the former respect, the intermediate step identifies the underlying technically beneficial realisation of the invention as a whole. This aligns with the general principles expressed by the Technical Board of Appeal in T1249/2022 at paragraphs 10 to 12 and T1769/10 at section 4.2.

What is the intermediate step for ANN patent claims?

The intermediate step is the analytical bridge between the any hardware exclusion threshold and the full novelty and inventive step assessment. It was described by the Enlarged Board of Appeal in G1/19 and endorsed by the Supreme Court as a necessary part of the analysis for computer implemented inventions. It acts as an objective filter.

For ANN based claims it requires a feature by feature examination of the claim to determine which elements contribute to technical character when the invention is considered as a whole. That is particularly important for mixed inventions, those combining ANN components with other system features, some of which may be technical and some of which may serve non technical purposes. The step asks two questions of each feature:

  1. Does the feature contribute to the technical character of the invention when considered contextually as a whole? This potentially eliminates technical features that are irrelevant to the invention.
  2. Does the feature, whether technical or not, interact with other technical subject matter of the claim in a way that supports delivery of an overall technical effect?

Features that pass, either as inherently technical contributing features or as non technical features that interact with the technical subject matter, are included in the inventive step assessment. Features that do not contribute to technical character are filtered out.

For applicants this means the specification and claims must clearly articulate the technical interactions between the ANN architecture, including how the ANN is trained and assessed relative to a cost function and ground truth, and other system components. An ANN training method that produces a specific technical effect, whether improved signal quality, greater compression efficiency or more accurate anomaly detection, through its interaction with hardware, sensors, networks or databases has a stronger claim to technical character than one whose described effects are purely functional or abstract.

How should AI developers structure ANN patent claims?

Claim the system, not just the model. A claim directed solely to a trained ANN, its weights, biases and topology, without reference to the hardware environment in which it operates, may struggle to clear the as such exclusion. A claim directed to a system that includes the ANN, hardware components and a defined method of operation is on stronger ground.

Describe the training method in technical terms. ANN training is a technical process. The choice of training objective, loss function, dataset structure and backpropagation configuration are technical decisions that can contribute to technical character. The specification should explain why those choices were made and what technical effect they achieve, not simply describe them as steps in a generic training loop. This supports protection of the ANN itself as a product by process.

Identify what the trained model does differently. The inventive step assessment will focus on what is different about the claimed invention compared with the prior art. For ANN inventions that often means explaining clearly what the trained model can do that prior art systems could not, with precision about the technical mechanism rather than the commercial outcome. The objective is identification of a new product produced by a new process to deliver a new result.

Preserve dependent claim fallbacks on architecture. Architecture choices, the number of layers, the type of activation functions, the connectivity pattern and the embedding dimension, can all be technically significant, but they need to be expressed in the context of applied technology rather than as an abstract idea. Dependent claims specifying these choices may provide fallback positions in prosecution and can support an inventive step argument. They should be expressed in terms of the effects they achieve and why the structure was chosen to support a particularly beneficial technical realisation.

How are different types of ANN treated?

The court treated ANNs as a category without distinguishing between architectures or training paradigms, so the reasoning applies broadly.

Feedforward networks and deep learning models. Standard supervised models trained by backpropagation fall squarely within the description. The weights and biases are computer programs, the implementing system involves hardware, and the claim is not excluded as such.

Recurrent networks and transformers. These involve more complex connectivity and, for transformers, attention mechanisms rather than traditional weight layers. The same analysis applies.

Generative models. GANs, VAEs, diffusion models and large language models are all ANN based. Patentability depends on the same analysis: technical character and technical effect, feature interaction at the intermediate step, and novelty and inventive step of the specific architecture, training approach or application.

Embedded and hardware implemented ANNs. Where an ANN is implemented in dedicated hardware, whether a neuromorphic chip or an ASIC designed for inference, the implementation does not change the legal character of the ANN. It remains a program for a computer. It does, however, make it even clearer that the claim involves hardware.

Can the ruling help with previously rejected applications?

Potentially, yes, but it depends on the date of the rejection and therefore on whether the application can be revived. Applications refused, or subject to an ongoing refusal, primarily on the basis of the Aerotel contribution with rider analysis may be recoverable in light of the changed framework. That approach, now rejected as unsound, put the cart before the horse, asking a backwards facing question of what the invention had contributed to the font of human knowledge.

The change in law does not automatically entitle a refused application to allowance and grant. Each case depends on its facts: the claim language, the way the technical contribution was presented, the procedural status of the application and the strength of the underlying invention. But where an application was refused under Aerotel in circumstances that would not have produced a refusal under the any hardware approach, and particularly under an intermediate step that properly considers the invention as a whole, there may be grounds to revisit the position.

If you have previously been advised that UK patent protection for an ANN based invention was not viable, that position is worth reviewing. For innovations whether pre filing or in active prosecution, Hepworth Browne advises on strategy, specification drafting and prosecution for ANN based and computer implemented inventions, reflecting Bruce Dearling’s work before the UK Supreme Court in the Emotional Perception case. See also our artificial intelligence practice page.

Frequently asked questions

Is a trained neural network model a patentable invention? The trained model viewed in isolation is a computer program and therefore potentially excluded. A system that includes the trained model, hardware and other technical components may be patentable depending on the technical character of the claimed invention as a whole and, fundamentally, whether it is differentiated from the prior art.

Can the training process be patented? A training method can potentially be patented if it produces a technically inventive outcome, such as a trained model with particular technical properties achieved through a technically non obvious training approach. The claim must include technical features that deliver a tangible benefit in the final network.

Does it matter whether the ANN runs on a GPU, CPU or dedicated chip? For the as such exclusion, no. The choice of hardware implementation is irrelevant to the legal character of the ANN. For the intermediate step and inventive step analysis the specific hardware is generally irrelevant unless the choice itself contributes to the overall technical character of the invention.

How does the ruling affect strategy at the EPO? The EPO already applied the any hardware approach, so the ruling primarily aligns UK practice with the existing EPO framework and the use of the intermediate step. The EPO, however, views ANNs as mathematical methods rather than computer programs. Both require a positive application of the underlying technology to overcome assertions that the claimed invention is abstract. For dual UK and EPO applicants, the convergence means prosecution strategies developed for the EPO are more directly applicable to UK prosecution than before. The UK approach to inventive step, the Pozzoli questions and particularly whether the differences are obvious, remains different from the problem and solution approach. Pozzoli is arguably not inhibited by hindsight reasoning that can distort the actual invention when a problem is formulated from identified technical differences between the claimed invention and the prior art.

How does the ruling affect other jurisdictions? The approach taken by the Unified Patent Court, following headnote 1 of its Court of Appeal decision UPC-CoA-0000901/2025, now also aligns with the UK Supreme Court decision, albeit that both the UPC and the EPO favour the problem and solution approach to inventive step. Whether countries such as Singapore follow the decision appears likely, at least to the extent that any hardware and the intermediate step are objectively sensible assessment mechanisms. Whether they adopt the restrictive definition of an ANN as a computer program is less evident, and more liberal views appear technically and commercially more likely, especially given the response of the Association of Singapore Patent Attorneys to the IPOS consultation on the relevance of the decision to Singapore law. In light of the effective codification of Aerotel into New Zealand patent law at section 11, the decision holding Aerotel to be unsound may bear on the continuing appropriateness of that test there. Other national states influenced by the UK Supreme Court may be subject to the same line of argument.

Get in touch

Tell us what you have made.

A first conversation costs nothing and usually saves money. Send us an outline and we will tell you honestly whether there is something worth protecting, and what it is likely to cost.

Step 1 of 4

Where you are

What stage are you at?

This tells us who should pick up your enquiry and how quickly.

What you need

Which right fits best?

A rough steer is fine. We will confirm the right route once we understand your invention.

The invention

Give us a short outline.

Keep it high level. Please do not send confidential detail before we confirm terms.

Where to reply

Who shall we get back to?

Please add your name
Please add a valid email address

Enquiry ready

Thank you. An attorney will review your outline and respond within one working day.