Last month The UK Supreme Court delivered a significant ruling for the AI sector, one that tackles a key question for innovators: “Can AI-driven systems be patented in the UK?”
The appeal focussed on a key aspect, Artificial Neural Networks (ANN), systems designed to recognise patterns and classify data, and how these fit within the framework of the UK’s current Patents Act (1977).
The outcome could provide long-awaited clarity on the patentability of modern AI technologies and for developers building machine-learning tools, the decision could reshape how intellectual property is protected.
So, what does the ruling mean and what steps should AI innovators be taking to protect their inventions going forward?
The Black Letter of the Law
Under the Patents Act 1977 a “program for a computer” is excluded from patent protection “as such”. In practice, this means that software on its own is not patentable if the invention merely relates to the execution of a computer program. However, a computer-implemented invention may still qualify for a patent where it produces a technical contribution beyond the program itself, such as solving a technical problem or improving the functioning of a system.
How did the Supreme Court get to its decision?
The case arose from an application by Emotional Perception AI, a company focused on creating technology that analyses human emotions, for a system that uses an ANN to recommend media files (such as music, video or text) based on their ability to generate similar emotional responses in users. The UK Intellectual Property Office initially rejected the application, finding that the invention was merely a “program for a computer…as such,” which is excluded from patentability.
The central issue before the Supreme Court was how AI-driven systems, particularly those implemented through neural networks, should be treated under the long-standing exclusion for computer programs.
Earlier decisions in the case had produced conflicting outcomes. The High Court found the invention potentially patentable, while the Court of Appeal reinstated the UKIPO’s refusal on the basis that the claimed ANN did not provide a sufficient technical contribution. The appeal provided the Supreme Court with its first opportunity to address directly the patentability of modern AI technologies.
In its judgement, the Supreme Court clarified that AI-related inventions should not automatically be excluded from patent protection simply because they are implemented in software. The Court held that artificial neural networks are capable of constituting patentable inventions and rejected aspects of the previous analytical framework used in UK case law, aligning the UK approach more closely with European Patent Office practice.
The Court concluded that even though the ANN was a programme for a computer, it also involves technical means because the ANN “can only be implemented on some form of computer hardware”, which was sufficient to show that the subject matter of the claims has technical character and is not a computer programme “as such”. Concluding that UKIPO was therefore wrong in refusing the patent application on the grounds that it did.
A turning point for AI patents
The Supreme Court’s decision is being received across the industry as a watershed moment for how the UK patent system treats artificial intelligence and the ruling could influence how the current law is interpreted and applied in future cases.
While the finer details will be tested in future cases, the outcome signals an environment where genuinely technical innovations stand a stronger chance of overcoming exclusions and securing protection.
For AI innovators, the message is pragmatic but encouraging. Patents remain available for AI that solve a real problem in a technical way, provided that applicants can evidence and explain that contribution with precision. The practical task now is to calibrate strategy, drafting and evidence so that applications meet the heightened expectations on technicality.
4 steps AI innovators should take to protect their IP
1. Reassess whether AI technologies may now be patentable
The decision confirms that inventions involving ANNs should not automatically be excluded as “computer programs as such” as explored within this article. AI developers should therefore revisit technologies that may previously have been dismissed as unpatentable software. Where an AI system produces a technical effect or solves a technical problem, it may now have stronger prospects of patent protection.
2. Draft patent applications to emphasise technical contribution
The key question remains whether the invention provides technical contribution beyond a mere program. Applicants should therefore ensure that patent specifications clearly explain how the AI system operates technically, the architecture of the model, and the technical problem it addresses. Framing the invention in technical rather than purely functional or commercial terms will be critical.
3. Review existing IP strategies for AI products
Businesses developing machine-learning tools should reassess their broader IP strategy. Some technologies that were previously protected in other ways may now be suitable for patent filings. A balanced strategy combining patents and confidential information protection may offer stronger long-term protection.
4. Act early to secure priority in a fast-moving field
The decision is likely to prompt increased patent activity in AI-related technologies. Innovators should therefore consider filing applications early to secure priority dates before competitors. Early engagement with patent solicitors/attorneys can also help ensure that applications are drafted in a way that aligns with the developing interpretation of AI patentability.
What next?
The decision marks an important development in how UK law approaches AI-driven inventions.
While the Supreme Court has now clarified the legal framework under the Patents Act 1977, the dispute itself is not quite over. The Court has sent the matter back to the UK Intellectual Property Office for reconsideration of the patent application in light of its judgment.
The outcome of that will be closely watched by AI innovators, as it may provide the first practical indication of how this will be applied in practice in the future.
The boundaries of patentability for AI will continue to evolve as new technologies emerge and further cases are tested under the framework of the Patents Act 1977. For innovators, the key takeaway is clear, AI systems should no longer be dismissed as automatically unpatentable software, and early consideration of IP protection will be critical.
