Artificial intelligence in drug discovery: the legal status of AI-generated inventions

Charlotte Tillett
Stevens & Bolton, Guildford
charlotte.tillett@stevens-bolton.com

Kate Hamson-Maguire
Stevens & Bolton, Guildford
kate.hamson-maguire@stevens-bolton.com

George Harrington
Stevens & Bolton, Guildford
george.harrington@stevens-bolton.com

The impact of AI on the drug development process

The development of new pharmaceutical products is an expensive and time-consuming process, often requiring inventors to test a large number of methods and formulae. Those who successfully develop new drugs will usually hope to obtain exclusive rights to financially exploit the product for a certain period of time. The patent system, governed in the United Kingdom by the Patents Act 1977, rewards inventors with the ability to generate profits and recoup R&D costs before competitors can enter the market with similar products.^[1]

The rapid development of artificial intelligence (AI) is likely to have a significant impact on this process. AI systems can test formulae and variations much faster than traditional methods, facilitating the development of new drugs within a shorter timeframe. For example, in 2022 DeepMind’s AI tool AlphaFold was used to predict and publish the structures of millions of proteins, a resource which scientists are already using to develop and test new drugs.^[2] Enhancements in the information available and the acceleration of testing processes may mean that pharmaceutical and biotech companies become less reliant on ensuring significant financial returns to offset their R&D costs, which in turn may encourage the development of less profitable (but much-needed) therapies and medicines.^[3]

Patentability of AI-generated inventions – the Thaler case

While AI offers significant promise for drug development, it also raises important questions about the legal status of inventions that are made with the assistance of AI or are generated solely by an AI system. The UK courts have recently addressed this in a dispute between the Intellectual Property Office (UK IPO) and Dr Stephen Thaler, although various questions remain unanswered.

Dr Thaler used his AI system, called the ‘Device for the Autonomous Bootstrapping of Unified Sentience’ (or DABUS), to generate two inventions: a container for storing food and a flashing beacon. Thaler then applied for UK patents to protect these inventions, naming DABUS as the inventor. When the UK IPO rejected his applications, Thaler commenced proceedings, arguing that DABUS should be recognised as an inventor and, as the owner of DABUS, he was entitled to ownership of the patents by virtue of the ‘doctrine of accession’.

The dispute reached the Supreme Court, which in 2023 ruled against Thaler. The Supreme Court referred to section 7 of the Patents Act 1977, which states that a patent may be given ‘primarily to the inventor’, defined as the ‘actual deviser of the invention’. Referring to existing case law (which established that an ‘inventor’ only covers natural persons), the Court decided that an artificial system cannot be an inventor, nor be granted a patent.^[4]

Thaler also filed an international patent application for the inventions and was given until June 2021 to identify the inventor. Thaler identified himself as the inventor in December 2023 (after the Supreme Court’s ruling), but by this point his application was deemed withdrawn as he had not complied with the deadline. Thaler appealed this decision, but in 2025 the UK Patents Court deemed his application to be ‘obviously defective’. Thaler had not complied with the relevant timeframes, and his application named himself as the inventor on some documents, but DABUS on others.^[5]

Implications of Thaler and the future of AI-generated inventions

The English courts’ approach aligns with the European Union and the United States, confirming that an AI cannot be considered an ‘inventor’ for patent purposes in the same way as a human.^[6] However, the Supreme Court’s judgment should not be interpreted as a conclusion on the patentability of AI-generated inventions generally. While Thaler disclosed that DABUS generated its products completely autonomously, in practice many inventions involve some use of AI in the development process, but a human will still be disclosed as the inventor.^[7] At this stage, the question remains open as to whether the use of AI will affect the patentability of an invention in future.

This is certainly not the first time that the law has had to grapple with the IP implications of AI-generated or AI-assisted works. The UK IPO has already held consultations addressing the interaction between AI and IP, and has (so far) concluded that the patent system is sufficiently equipped to address AI technologies.^[8] However, as AI systems become increasingly capable, the existing patent rules are likely to be tested by cases where the human inventor guiding the process is harder to define. For now, those using AI in drug development and clinical trials should ensure they can demonstrate meaningful human involvement in the process, and remain alert to any proposed legal changes, to ensure their inventions are not excluded from patentability on that basis.