UK Supreme Court overturns Aerotel in Emotional Perception AI patent case

Major shift in UK patent law interpretation aligns with European Patent Office approach
The UK Supreme Court has fundamentally reshaped the landscape for patenting artificial intelligence inventions, overturning the long-standing Aerotel framework and aligning UK law with European Patent Office jurisprudence. The unanimous judgement in Emotional Perception AI Limited v Comptroller General of Patents, Designs and Trade Marks [2026] UKSC 3, delivered on 11 February 2026, addresses whether artificial neural networks constitute computer programmes and how computer-implemented inventions should be assessed for patentability.
Emotional Perception AI sought to patent a system using an artificial neural network to recommend media files based on emotional similarity. The invention trains an ANN to correlate physical properties of files (such as tempo and rhythm in music) with semantic descriptions of human emotional responses. After training, the system recommends content likely to elicit similar emotional responses based solely on measurable physical characteristics.
The application was initially refused by the UK Intellectual Property Office under the computer programme exclusion in section 1(2)(c) of the Patents Act 1977, which implements Article 52 of the European Patent Convention. Sir Anthony Mann allowed the applicant's appeal, finding that ANNs are not computer programmes. The Court of Appeal reversed this decision, but the Supreme Court has now charted a different course entirely.
Lord Briggs and Lord Leggatt, delivering the leading judgement with which the other Justices agreed, held that the Aerotel methodology should no longer be followed. That approach, established in 2006, required courts to identify the "actual contribution" made by an invention and assess whether it was technical in nature. The Supreme Court accepted that this was fundamentally incompatible with the interpretation of Article 52 adopted by the EPO's Enlarged Board of Appeal in decision G1/19 (2021).
The Court endorsed the EPO's "any hardware" approach as the first stage of analysis. If a claim involves technical means—such as a computer—it avoids exclusion under Article 52(2) and constitutes an invention. However, this creates only a low threshold. A crucial intermediate step must then identify which features contribute to the technical character of the invention as a whole, filtering out non-technical features before assessing novelty and inventive step.
On the central question of whether ANNs constitute computer programmes, the Court sided with the Comptroller General. The judgement clarifies that an ANN is an abstract mathematical model, not a physical machine. Whether implemented on dedicated hardware or a conventional computer, an ANN represents a set of instructions for manipulating data in a particular way. The network's topology, activation functions, weights and biases collectively instruct the machine to process data, making the ANN as a whole a programme for a computer.
Nevertheless, the claimed invention is not a computer programme "as such" because it necessarily involves hardware implementation. The Supreme Court allowed the appeal and remitted the case to the UKIPO to assess novelty and inventive step applying the correct legal framework.
The Court declined to provide comprehensive guidance on applying the intermediate step, noting this would require detailed argument not presented in the appeal. The judgement acknowledges the potentially disruptive effect on established practice but emphasises that consistency with EPO jurisprudence on treaty interpretation is paramount. Importantly, the Court confirmed that adopting the EPO's interpretation of Article 52 does not require abandoning the Pozzoli approach to inventive step, which remains good law following Actavis v ICOS.
This decision marks a watershed moment for UK patent law, particularly for AI and computer-implemented inventions, whilst leaving key questions about practical application for future determination.
