CLA News / AI and Patents – The Thaler Judgment and The Question of Inventorship By Fadil Syed Hashmey
Artificial intelligence (“AI”) refers to a machine-based system capable of generating advanced responses based on data inputs. It has an increasingly significant presence across the scientific and technological sectors and can perform a wide range of functions autonomously.
This article analyses the growing importance of AI in the modern era. It examines the ‘inventor’ requirement under the UK Patents Act 1977 (the “Act”), which has been interpreted to exclude the possibility of AI being recognized as an inventor in terms of patent protection.
It further considers the issues addressed by the UK Supreme Court (“UKSC”) in Thaler v Comptroller General of Patents[1] (“Thaler“).
Thaler stems from two patent applications made to the UKIPO by Dr Thaler concerning products that were created by a sentient AI named DABUS.[2] The Comptroller-General of Patents (“Comptroller”) rejected Dr Thaler’s patent applications as DABUS could not qualify as an ‘inventor’ for the purposes of the Act. After losing by majority ruling in the Court of Appeal (“CoA”),[3] Dr Thaler appealed to the UKSC. The core question before the UKSC was whether a sentient AI could be named as an inventor of a patent under the Act.
This article evaluates the UKSC’s findings, and examines the views of legal academics and practitioners, many of whom agree with the UKSC’s decision but remain concerned about the wider policy implications.
Finally, the article concludes that, although the UKSC was legally correct in adopting a textualist approach, Parliament should amend the Act to specifically address the evolution in AI-related patent protection.
What is AI?
‘AI’ has not been defined in any UK statute. However, the UK Government has described AI as “technologies with the ability to perform tasks that would otherwise require human intelligence, such as visual perception, speech recognition, and language translation.”[4] Similarly, the EU Artificial Intelligence Act[5] defines AI as “a machine-based system” that can generate “outputs such as predictions, content, recommendations, or decisions that can operate autonomously and influence physical or virtual environments”.
1.1 What constitutes a patentable invention and what is the use of AI in the generation of patentable inventions?
The World Intellectual Property Organization (“WIPO”) broadly defines an ‘invention’ as “a new and inventive solution to a technical problem.”[6] A patent, as also defined by the WIPO,[7] is a type of exclusive, monopolistic, and registerable intellectual property right that prevents others from exploiting an inventor’s creation without the inventor’s permission.
For an invention to be patentable in the UK, the Act requires that it be novel, involve an inventive step, and be capable of industrial application.[8] This wording in the Act is derived from the European Patents Convention (“EPC”).[9]
In terms of general patentable inventions, AI has permeated the modern-day workspace. The WIPO has reported that ‘image-recognition’ technology has been used in 49% of AI-related patent applications[10], which, in turn, have been utilized for security[11] and navigation purposes,[12] with further patent applications for language and speech processing[13] following closely behind.
Thus, the integration of AI in the generation of patentable inventions has assisted in harmonizing the research and design process, leading to more efficient and cost-effective solutions.[14]
1.2 The development of the relationship between patents and artificial intelligence
The AI boom has brought instrumental changes in the patent process. Apart from the essential requirements for a valid patent under Sections 1(1) and 1(2) of the Act, the UKIPO, issued a separate guideline[15] (“Guideline”) distinguishing which AI-related inventions can be patented. Although the Guideline was cited in Thaler, it was subsequently withdrawn following the recent case of Emotional Perception[16] in favor of an approach that is more harmonious with the European Patents Office (“EPO”).
Section 1(2) expressly excludes certain categories from patentability, but Guideline 4.10 suggested that AI inventions could avoid exclusion if they had made an appropriate ‘technical contribution.’[17] The courts[18] had also developed signposts for determining whether an AI invention satisfied the ‘technical effect’ criteria.[19] These signposts were not definitive, but the presence of a ‘technical effect’ indicated a broader ‘technical contribution’.
1.3 What constitutes an inventor and who can be named the inventor of a patent?
In the UK, there is a core requirement that only natural ‘persons’ can be legally recognized as inventors for the purpose of obtaining a patent.[20] This reasoning has been amplified by the courts through recent decisions.[21]
The Act, itself, does not expressly distinguish between ‘AI-assisted’ inventions and ‘AI-generated’ counterparts. However, a distinction may be drawn in order to better outline the rules on inventorship. AI-assisted inventions involve the presence of a human inventor where AI assists in the data analysis or research process, with the human inventor retaining a central role in devising the invention. Contrarily, AI-generated inventions are autonomously created by AI, with minimal or no human input. They are subject to stricter requirements in terms of patentability[22] as confirmed by the UKSC in Thaler.
Under Section 7(3) of the Act, as long as a human can be legally identified as the inventor, utilizing AI as a ‘medium’ for human innovation appears permissible and consistent with the concept of an ‘actual deviser’. On its own, however, AI cannot occupy the legal position of an inventor.
2. Key issues addressed by the UKSC in the Thaler judgment
Thaler is instrumental in delineating the relationship between AI and patent ownership under the Act.
The UKSC addressed three main issues, namely: the meaning of the term ‘inventor’, who is entitled to make an application for a patent and to whom may a patent be granted,[23] and whether the hearing officer was entitled to withdraw the application.
2.1 Issue (i) – The meaning of an ‘inventor’ under the Patents Act 1977
In assessing the definition of an ‘inventor’ the UKSC adopted a textualist approach. The ordinary meaning of Section 7(3) of the Act refers to an ‘inventor’ as the ‘actual deviser’ of the invention. Thus, Lord Kitchin was of the view that an ‘inventor’ can only be a natural ‘person’.
The interpretation of the word ‘person’ in the context of patents was previously explored in Yeda Research.[24] In Yeda, Lord Hoffman relied on Laddie J’s finding in University of Southampton’s Applications.[25] He concluded that the definition of ‘inventor’ is constrained to “the natural person who came up with the inventive concept.”[26] Through this interpretation, the word ‘natural’ impliedly excludes AI.
Section 7(2) of the Act identifies persons to whom a patent may be granted, providing that a patent can be granted to “no other person”. This wording supports the idea that an inventor must be a natural person. Section 13 describes the right of the inventor to be identified in the patent application, reinforcing the idea that an inventor must be a person.[27]
Therefore, the UKSC decided that the title of ‘inventor’ could not plausibly be extended to a non-natural person without a legal personality.[28]
2.2 Issue (ii) – Was Thaler the owner of any invention made by DABUS and was he entitled to obtain a patent in respect of it?
The UKSC described Section 7 of the Act as a “complete code” for determining the entitlement to apply for, and receive, patents.[29] Accordingly, Dr Thaler’s first hurdle came from the fact that he was not the primary or joint ‘inventor’[30] or ‘actual deviser’ of the patents, as required under Section 7(2)(a). Similarly, DABUS was not a natural ‘person’.
The UKSC refuted Dr Thaler’s argument on entitlement to the patent under Section 7(2)(b)[31] through the common law doctrine of accession. Lord Kitchin distinguished that the doctrine of accession concerns “new tangible property produced by existing tangible property” and is not extendable to intangible concepts such as patents.[32]
The UKSC opined that Dr Thaler was not entitled to file for, or be granted the patents, given that DABUS could not be identified as a ‘person’ within the purview of the Act. The UKSC also relied on the point that Dr Thaler had made no technical advances towards the creation of the products, and therefore had no independent right to obtain the patents.[33]
Patten and Zapalowski[34] agree with the UKSC’s decision, and emphasize that Parliament must be the source of any change in the law. They also allude to the UKIPO’s proposition for introducing AI-specific protections,[35] which purported to provide rights comparable to a patent, but with lesser exclusivity, as being a safe ground for AI patents. However, these proposals were not implemented.
2.3 Issue (iii) – Was the Hearing Officer entitled to withdraw the patent applications?
The express language of Section 13(2)(a) places a duty on the applicant to mention the ‘person’ who is believed to be the ‘inventor’ of the patent.
The UKSC reiterated that identifying a non-person, DABUS, as the inventor, could not legally fulfil the requirements of Section 13(2)(a).[36] For the same reasoning, Dr Thaler’s argument under Section 13(2)(b) also failed as there was no natural inventor from whom Dr Thaler could derive his rights to the patent, thereby resulting in the automatic withdrawal of the application.[37] The UKSC cited Nippon Piston Ring[38] as support for the notion that withdrawal of the application was a statutory consequence mandated by Section 13(2).
- How has the Thaler decision has been received in legal articles?
The reception to Thaler has been mixed, with most commentators agreeing with the ratio decidendi, while others argue that the refusal to recognize sentient AI as ‘inventors’ has stifled industrial progress.
Saravanan and Deva[39] suggest that the court’s cautious approach towards AI patentability may reduce the incentive for creators to use AI. This, in turn, would make for an inefficient research and design process.
Matulionyte[40] states the UKSC’s decision retains human involvement in the creation process. She acknowledges that if such restrictions to AI-based patentability were not present, it would open a floodgate of AI patents. She suggests that since large corporations possess the most sophisticated AI tools, this would make it difficult for smaller companies to commercialize new technology, leading to a ‘concentration’ effect. Matulionyte further opines that the law currently incentivizes companies to credit human inventors for inventions, even where AI is the creator, thereby undermining the fairness and transparency of the patent system.
Smartt[41] also believes that the Act is in dire need of an update, outlining how the definitions provided in the Act date back to the 1970s, when Parliament did not have sentient AI in mind.
Ryan Abbott,[42] who represented Dr Thaler before the UKSC, contends that denying patent protection for AI inventions forces an overreliance on trade secrets as a means of protection. He urges Parliament to amend the Act in view of the increasing financial implications of AI-generated products.
- Implications of the Thaler decision and evaluation of UKSC’s approach and decision
Thaler comes as an obstacle for any sentient AI-based patent applications. The UKSC’s decision is legally sound and technically correct. However, it exposes the perforations within the Act and highlights the disconnect between the law and technological realities.
The UKSC’s textualist approach concluded that an ‘inventor’ under the Act cannot mean non-natural persons, as Parliament did not intend to include machines.[43] Therefore, the decision suggests that if a company cannot identify a human being as an “actual deviser” in their patent applications, they may be inclined to explore other options.
Firstly, they may avoid investing in AI tools that cannot provide enforceable IP protection rights. This option may be unpopular for sectors that are highly competitive and depend on research innovations.
Secondly, companies may resort to using trade secrets as a means of protection, which would reduce transparency and collaborative research.
Finally, they may simply declare a human as the inventor in patent applications, despite the substantial work having been carried out by AI. The commercial incentive indicates that this is likely to be the most exploited alternative.
- Conclusion and Reform
From a policy-centric perspective, Birss LJ’s dissent in the CoA reveals the depths of the currently fractured legislation. His observation that if Thaler had named himself as the inventor “none of the problems would arise”,[44] suggests that factually misleading the UKIPO cannot be punished or remedied under the current statute. This further indicates the insufficiency of the Comptroller’s powers to review a patent based on the possibility of AI involvement, suggesting that a ‘human-centric’ model of inventorship is no longer suitable.
However, Birss LJ’s observation carves open an additional aspect of the debate, specifically whether a mandatory disclosure of AI involvement ought to be made when filing a patent application. Such disclosure appears be a step in the right direction, and AI-usage detection services can also be employed to tighten the vetting process. Similarly, the Act ought to provide for a clear distinction between AI-assisted and AI-generated patents, encouraging technological innovation while maintaining inventor integrity.
The UKIPO’s stance on Thaler is not incorrect – the USA,[45] Australia,[46] New Zealand[47] the EPO[48] have all rejected the concept of AI as a legal inventor under their respective statutes. Thaler hammers away at the limits of a framework that is increasingly disconnected from the technical realities of the modern day. Therefore, Parliament should consider introducing a timely amendment to the Act to prevent the UKIPO from being swept away by an uncontrollable and unchecked wave of AI-driven innovation.
Fadil Syed Hashmey
Fadil Syed Hashmey is a Solicitor of the Senior Courts of England and Wales and an Advocate of the High Courts of Pakistan. He holds an LL.M. in Legal Practice from the University of Law. He is based in the United Kingdom.
[1] ([2023] UKSC 49)
[2] DABUS <https://imagination-engines.com/dabus.html> accessed 2nd August, 2025
[3] Thaler v Comptroller General of Patents, Designs and Trade Marks [2021] EWCA Civ 1374
[4] Guideline 3, Guidance for examining patent applications relating to artificial intelligence <https://www.gov.uk/government/publications/examining-patent-applications-relating-to-artificial-intelligence-ai-inventions/guidelines-for-examining-patent-applications-relating-to-artificial-intelligence-ai–2#:~:text=AI%20inventions%20are%20computer%2Dimplemented,a%20program%20for%20a%20computer> accessed 30th July 2025
[5] Article 3(1) of the Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 <https://artificialintelligenceact.eu/article/3/> accessed 21st July 2025
[6] WIPO, ‘Inventing the Future’ Business Series Number 3 <https://www.wipo.int/edocs/pubdocs/en/wipo-pub-917-1-en-inventing-the-future.pdf> accessed 30th July 2025
[7] WIPO, ‘Patents’ <https://www.wipo.int/en/web/patents> accessed 30th July 2025
[8] Section 1(1) Patents Act 1977
[9] Article 52(1) European Patents Convention
[10] WIPO, “The Story of AI in Patents” <https://www.wipo.int/tech_trends/en/artificial_intelligence/story.html> accessed 5th August 2025
[11] West and Allen (n 11) “AI can process large datasets such as camera and video surveillance, which it can, in turn, used as a baseline to detect suspicious or abnormal activity”
[12] Airbus, “Implementing AI throughout our business” <https://www.airbus.com/en/innovation/digital-transformation/artificial-intelligence> accessed 5th August 2025
[13] Lugbara, Luganda and Ancholi, “Using machine learning to analyse radio talk in Uganda” UN Global Pulse 2017, “In Uganda, a project used machine learning to convert radio broadcasts into text that could be translated and read in the various languages spoken in Uganda” <https://unsdg.un.org/sites/default/files/Using-machine-learning-radio-content-uganda.pdf> accessed 5th August 2025
[14] PwC, “Sizing the Prize. What’s the Real Value of AI for Your Business and How Can You Capitalise?”, A project by PwC has estimated that “artificial intelligence technologies could increase global GDP by $15.7 trillion, a full 14%, by 2030.” (2020). <https://www.pwc.com/gx/en/issues/analytics/assets/pwc-ai-analysis-sizing-the-prize-report.pdf> accessed 30th July 2024
[15] UK IPO, “Guidelines for Examining Patent Applications Relating to Artificial Intelligence (AI)” GOV.UK, (30th January, 2025) <https://www.gov.uk/government/publications/examining-patent-applications-relating-to-artificial-intelligence-ai-inventions/guidelines-for-examining-patent-applications-relating-to-artificial-intelligence-ai–2> accessed 30th July, 2025
[16] Emotional Perception AI Ltd v Comptroller-General of Patents [2026] UKSC 3
[17] The ‘technical contribution’ aspect was derived from the four-step approach in Aerotel Ltd. v Telco Holdings Ltd [2006] EWHC 997, particularly relevant for computer programs and mathematical methods which were excluded from patentability under Section 1(2)
[18] Relevant cases include: Merrill Lynch Inc’s Patent Application [1989] 4 WLUK 219, HTC v Apple [2013] EWCA Civ 451, and Symbian [2008] EWCA Civ 1066
[19] Guideline 4.14 and 4.16 (n 15)
[20] Section 7 Patent Act 1977
[21] Richard Ellis, “Protecting Your Invention: Understanding the Impact of AI on Patients” Freeths, <https://www.freeths.co.uk/insights-events/legal-articles/2025/protecting-your-invention-understanding-the-impact-of-ai-on-patents/> accessed 1st August, 2025
[22] “The Interaction between Intellectual Property Laws and AI: Opportunities and Challenges” Norton Rose Fulbright <https://www.nortonrosefulbright.com/en/knowledge/publications/c6d47e6f/the-interaction-between-intellectual-property-laws-and-ai-opportunities-and-challenges#7 > accessed 1st August, 2025
[23] Thaler (n 28) [25]
[24] Rhone-Poulenc Rorer International Holdings Inc v Yeda Research and Development Co Ltd, [2007] UKHL 43
[25] University of Southampton’s Applications, [2004] EWHC 2107 (Pat)
[26] Yeda (n 32) [20], Thaler (n 28) [63]
[27] Thaler (n 28) [65]
[28] Ibid, “This machinery is again consistent with the proposition that rights are derived through the inventor and that patents are only granted or transferred to persons with legal personality.
[29] Thaler (n 28) [75]
[30] As defined under Section 7(3) Patents Act 1977
[31] Thaler (n 28) [81], This category typically applies in situations such as inventions made by employees, where Section 39 (1) of the 1977 Act stipulates that the invention belongs to the employer
[32] Thaler (n 28) [88]
[33] Thaler (n 28) [75], [84]
[34] J Patten and A Zapalowski, ‘AI is persona non grata: Supreme Court holds that AI is not an inventor’ (2024) 35(3) Ent. L.R. 133
[35] Intellectual Property Office, Artificial Intelligence and IP: Copyright and Patents: Government Response on Consultation (28th June 2022) <https://www.gov.uk/government/consultations/artificial-intelligence-and-ip-copyright-and-patents> accessed 23 August 2026
[36] Thaler (n 28) [94]
[37] Thaler (n 28) [98]
[38] Nippon Piston Ring Co. Ltd.’s Applications [1987] R.P.C. 120, Whitford J “It has long been a requirement in connection with the grant of letters patent that the name or names of the inventors should be identified. The 1977 Act, in section 13(1), merely repeats requirements which are to be found in earlier Acts with regard to the right of the inventor or inventors to be named in patents granted upon inventions which they have made”
[39] Saravanan A. and Deva Prasad M, “AI as an Inventor Debate under the Patent Law: A Post-DABUS Comparative Analysis” (1st August, 2024). European Intellectual Property Review, Volume 47 (1). pp. 26-39 SSRN: <https://ssrn.com/abstract=5053108> accessed 3rd August, 2025
[40] Rita Matulionyte, “‘AI Is Not an Inventor’: Thaler v Comptroller of Patents, Designs and Trademarks and the Patentability of AI Inventions” (2024) 88 The Modern Law Review 205 <https://doi.org/10.1111/1468-2230.12907> accessed 4th August 2025
[41] U Smartt, “Can Robots Have Feelings? Should We Now Apologise to the AI-Beast Called DABUS and Compliment ANNs Instead?” (2024) 46(3) European Intellectual Property Review 183, 184
[42] R Abbott, “Reflections on ‘Thaler v Comptroller-General’ and the Use of Artificial Intelligence in Research and Development” (2024) 46 European Intellectual Property Review 322
[43] Thaler (n 28) [79]
[44] Thaler (n 30) [81]
[45] Thaler v Vidal, 43 F.4th 1207 (Fed Cir 2022)
[46] Commissioner of Patents v Thaler [2022] FCAFC 62
[47] Thaler v Commissioner of Patents [2023] NZHC 554
[48] J 8/20 and J 9/20
