Emotional Perception 2026: the UK ruling that reshapes AI patent rules and the role of translation

11 February 2026 is a date that IP practitioners will remember. The UK Supreme Court issued its Emotional Perception judgment, significantly reshaping the criteria for patentability of software and artificial-intelligence-based inventions. For IP firms managing international filings, the key issue is not only the legal update itself, but understanding how this shift may affect the drafting and interpretation of patent texts across different jurisdictions.

The case Emotional Perception in brief

The UK Supreme Court delivered its long-awaited judgment in Emotional Perception AI v Comptroller General of Patents [2026] UKSC 3, concerning the patentability of artificial neural networks. The decision overturns decades of UK case law and aligns the country’s approach with that of the European Patent Office (EPO), abandoning the long-standing Aerotel test.

Emotional Perception AI had developed a system based on an artificial neural network capable of analysing the physical characteristics of multimedia files, including features such as frequency, rhythm and timbre, and mapping them to emotional states in order to recommend content with similar physical properties. This approach goes beyond traditional genre-based classification.

The UK Intellectual Property Office initially rejected the application, arguing that the invention was a “computer program as such” and therefore excluded from patentability. The case had already passed through several levels of appeal before reaching the Supreme Court, which ultimately overturned the earlier decisions and recognised the invention as patentable.

To understand the significance of this ruling, however, it is necessary to take a step back.

Farewell to the Aerotel test

Under the Aerotel test, in force since 2006, courts and patent examiners assessed software patent applications through four steps: interpreting the claim properly, identifying the actual contribution of the invention, determining whether that contribution fell within an excluded category, and finally assessing whether the contribution had a technical character.

Over the past twenty years, this approach has often been perceived as more rigid than the practice followed by the EPO.

In the Emotional Perception case, the Supreme Court adopted the “any hardware” approach developed by the EPO. Under this framework, a claim is not excluded from patentability if it incorporates or requires the use of any physical hardware component, even the most ordinary.

The EPO then assesses whether the invention produces a “further technical effect” beyond the normal interaction between software and hardware. If it does, the invention may qualify for patent protection even when it includes software.

With the Emotional Perception ruling, the UK therefore moves closer to the European framework and adopts a more substantive analysis of the technical contribution. This realignment is positive in terms of harmonising approaches between the two systems, but it also introduces a phase of interpretative transition. Transition phases are precisely when language becomes a critical factor.

Neural network, software, or technical system? The power of categories

One of the most delicate aspects of the judgment concerns the classification of neural networks and machine learning models.

The Court held that a “program” should be understood as a set of instructions executable by a computer to produce the desired data manipulations, and that an artificial neural network falls within this definition. However, this does not automatically exclude it from patentability. It is still necessary to determine whether the invention includes technical features that go beyond merely being a computer program, which is the category excluded from patentability by law.

Defining a neural network as a “computer program” or as a “technical system” is not a neutral choice. In patent law, legal categories determine which exclusions apply.

If the neural network is characterised as a pure program, it falls into the excluded category. If it is characterised as a technical system producing real-world effects, it may qualify for patent protection.

Terminology therefore becomes a strategic decision. Depending on the formulation adopted:

a formulation focused on technical contribution strengthens the patentability of the invention because it shifts attention to the effect produced in the physical world
a formulation that emphasises algorithmic abstraction brings the invention closer to the category of computer programs as such and exposes it to the risk of exclusion

In the post-Emotional Perception UK context, where the system seeks a balance between formal exclusions and substantive technical contribution, terminological consistency between the original version and its translations becomes even more critical.

Multi-country filings: the risk is not the error, but the nuance

Companies will not necessarily need to prepare different types of applications for the UK and for Europe. This alignment represents an important simplification for those managing multi-jurisdictional patent portfolios.

However, even between European and UK filings, words do not always carry the same weight.

Terms such as:

“algorithm” and “model”
“engine” and “technical system”
“computer-implemented method”

may carry very different implications depending on the regulatory framework and examination practice.

For IP firms, the real risk is not an obvious error but rather a subtle divergence between language versions that only emerges during examination or, worse, during litigation.

The shift toward the EPO approach is not a complete adoption. The analysis of inventive step, the requirement that an invention represent a non-obvious advancement over the state of the art, remains within the competence of UK courts and may involve adjustments compared with European practice.

When the interpretative framework evolves, textual stability becomes a strategic asset.

Context-aware translation in the IP field

A translator specialising in intellectual property does not intervene in strategic protection decisions. The translator’s role is to ensure that the terminological choices already made by the IP advisor remain intact when moving from one language to another and from one legal system to another.

In other words, the chosen term must correspond, in translation, to an equivalent legal category rather than a more vulnerable one.

To achieve this, it is essential to:

analyse terminology in light of the technical and patent context, not only the linguistic one
maintain terminological consistency across entire portfolios, ensuring that the same concept is rendered consistently in every document
monitor the evolution of legal categories that may affect textual interpretation, as happened with this ruling
flag potential terminological risks that could affect the robustness of a patent during examination or litigation

At a time when the UK is redefining the balance between software exclusions and technical contribution, linguistic precision is not a mere editorial detail. It is part of risk management.

Conclusion

The Emotional Perception decision is a concrete example of how legal categories evolve over time. When interpretative rules change, words acquire even greater weight.

For IP firms operating internationally, patent translation is not the final step in the process but a structural element of the protection strategy. Those managing multi-jurisdictional patent portfolios need partners who understand the technical and legal context in which they operate and who can ensure the linguistic consistency of patent texts over time.