Skip to content 
Artificial intelligence is transforming the way organisations function from automating customer service to creating completely new goods and procedures. But that raises an increasingly urgent question: if an AI system generates something new, can you patent it? And if you establish a firm on an AI-powered innovation, are you truly safe?
The answer – at least in present UK law – is complicated. These are exactly the problems recent court rulings have been wrestling with, and the answers are hugely important for business owners and entrepreneurs investing in AI-based innovation. In layman’s terms, this article explains the important cases and what it means to you.
First Things First: What is a Patent? Why Does It Matter?
A patent gives you the right to prevent anyone from creating, using or selling your innovation for up to 20 years. Patents are a potent commercial weapon for enterprises, enabling you to defend your market share, license your technology to others for revenue and dissuade competitors from imitating you.
In the UK, patents are regulated by the Patents Act 1977 and administered by the Intellectual Property Office (IPO). But not everything is patentable. Among other things, computer programs and mental acts are expressly excluded from patent protection. This exclusion has generated considerable difficulty in the field of AI and software-based inventions.
“The question courts keep asking is, does your AI-driven invention make a real-world, technical contribution — or is it just software doing what software does?
The Aerotel Framework: The Four-Step Test for Every AI Patent
UK courts’ assessment of patents involving AI is initially referenced in the case of Aerotel Ltd v Telco Holdings Ltd [2007] RPC 7. The Court of Appeal determined the case and set out a structured four-stage process that patent examiners and judges had to go through whenever an application incorporates potentially prohibited subject matter such as software or computer programs.
To summarise, the four steps are:
- Correctly interpret what the patent claim truly states.
- Find out what the real-world impact of the invention is — what has it really added to the world?
• Ascertain whether that contribution lies wholly inside the excluded categories (e.g., a pure computer programme or mental act).
- Consider whether it does offer a meaningful technical contribution even if it appears to fall inside exclusions.
For business owners, the message from Aerotel is this: Just because your idea runs on software or uses AI doesn’t automatically mean it’s not patentable. The question is if there is a meaningful technical influence in the world beyond the software itself. If your AI system manages a physical manufacturing process more efficiently or analyses medical imaging, leading to better clinical results, then that technological effect could be patentable.
The challenge is when the innovation is truly the program logic itself – a smart algorithm, a new way of arranging data, or a novel AI model – with no obvious technological effect outside the computer. That, the courts have always said, isn’t something patentable under UK law.
AT&T Signposts: 5 Questions to Ask About Your AI Invention
The concept was developed further on the basis of Aerotel in AT&T Knowledge Ventures v Comptroller of Patents [2009] FSR 19. Mr Justice Lewison created five ‘signposts’ – practical pointers that are helpful in deciding whether an innovation represents a real technical contribution. These signposts have now become a major part of the IPO and courts’ evaluation of AI and software patent applications.
In practice the five signposts are questions:
- Does the invention have a technological effect that extends beyond the regular physical interactions of the computer running the program?
- Does it have a technical effect upon a process outside the computer (for example, controlling machinery or developing a communications network)?
- Does it fix a technological problem inside the computer itself, such as running more efficiently or processing data more quickly?
• Is the contribution from a new sort of programme, not merely a new programme doing the same old things?
- What are the patentability implications if the contribution were to be implemented in hardware instead of software?
This last signpost is very helpful as a mental exam for entrepreneurs. Pull out the programme and ask, ‘If this idea were in a tangible machine or circuit, would it be patentable?’ If the answer is yes, then there is a convincing argument to say your AI-generated version is entitled to the same protection.
Tip for Business Owners: Before investing in a patent application for an AI or software-based breakthrough, ask your IP solicitor to map your idea against these five signposts. This will provide you with a realistic picture of your prospects before you invest money in the application procedure.
The Game-Changing Case: Emotional Perception AI Ltd v Comptroller [2023]
The most important and recent advance in this field was in the case of Emotional Perception AI Ltd v Comptroller-General of Patents [2023] EWHC 2948 (Ch), decided by Mr Justice Mann in the High Court. This ruling has sent shockwaves through the IP industry and truly changes the environment for organisations who work with AI – particularly those who are using machine learning and neural networks.
Emotional Perception AI has built a system that uses an artificial neural network (ANN) to propose music files to users based on semantic similarity — effectively recommending songs that ‘feel’ similar to one another, rather than just matching genre or artist tags. The IPO denied the patent because it was a computer program as such and was therefore not patentable.
Mr Justice Mann had a different view. His reasoning is fascinating and worth understanding since it fundamentally questions how we categorise AI systems under patent law. “The judge held that a trained neural network is not just a computer program in the conventional sense. ‘ Once trained, it works more like a custom-built machine and the rules the network follows aren’t created by a human programmer but come from the training process itself.’
That divergence is hugely significant. Traditional software exclusions were predicated on the notion of a traditional computer program — code produced by humans according to a set of explicit logical instructions. The judge thought a neural network was different in kind. The internal “rules” of the network are not human-written but learned from the patterns in the data after training. It may be inappropriate in principle to consider such a system as just a ‘computer programme as such’.
The judge further found that the system had a technological effect in that it did, in fact, cause certain files to be recommended and transferred. Even if the final user experience was the emotional or semantic quality of the advice, the method behind it was technological.
What does this mean for your business?
And if you’re producing a product or service that depends on machine learning, neural networks or other AI technologies, the Emotional Perception scenario is potentially extremely good news. That means AI-driven innovations may have a better claim to patent protection than previously thought—at least where the AI system itself is innovative and generates a real-world output.
But there are provisos. This is a first-instance High Court decision, and the IPO has said that it will continue to adopt the Aerotel framework till the wider law is resolved. It is possible that the case will be appealed or that further instructions may come. The law here is really fluid.
Here’s what we recommend for business owners at the moment:
• Don’t take it for granted that your AI innovation is automatically unpatentable because it involves software or machine learning. The law has progressed and may progress farther.
- Don’t assume it is patentable automatically either. The question of technical contribution remains important, and poor submissions will continue to be rejected.
• Document your innovation thoroughly – how the AI system has been taught, what data has been used, and what technical outputs it provides. If you ever decide to file a patent application, this documentation will be crucial for you.
- Consider a Freedom to Operate analysis together with any patent strategy. Even if you can’t patent every part of your AI system, it’s critical to know what patents have been filed by others in your sector.
• consult specialist advice promptly. AI patent strategy is still fast evolving, and generic advice from non-specialists can be costly.
What about AI as an inventor? The DABUS Conundrum
One extra dimension is worth highlighting. All of the preceding cases are human inventors using AI as a tool. Another subject of interest from a legal point of view is, may an AI system be listed as the inventor of a patent?
This was recently considered by the UK Supreme Court in Thaler v Comptroller-General of Patents [2023], which held that under existing UK legislation the creator must be a human being. The inventor has to be a person, and a patent application with an AI inventor alone will be disallowed. This has important ramifications for companies where AI is a fundamental generative aspect of the creative process: the humans who are managing, guiding and deploying the AI system need to be traceable and nameable as innovators.
This is an area where the law may develop in the future and where international variances are already becoming apparent. Companies with global IP strategies should watch changes in the US, EU and other relevant jurisdictions closely.
How Lawdit Can Help
Lawdit Solicitors has been helping firms with intellectual property strategies for over two decades. We know that for many organisations AI is not a future wish list item – it’s a present reality, and protecting the competitive advantage it offers is crucial.
Whether you’re a start-up building a machine learning platform, an established company automating operations with AI, or an entrepreneur with a really new AI-driven product, we can assist you in navigating this fast-moving terrain.
We provide patent searching and freedom to operate advice, drafting and prosecution of patent applications before the UK IPO and the European Patent Office, IP audits to identify and protect your most valuable innovations, and strategic advice on how to build an IP portfolio that supports your commercial goals.
info@lawdit.co.uk
