By Deepti Chaturvedi
In 2021, a landmark case in the United Kingdom (UK) raised fundamental questions about the role of artificial intelligence (AI) in intellectual property (IP) law: Dr. Stephen L. Thaler (Appellant) v Comptroller-General of Patents, Designs and Trademarks (Respondent). The case concerned an AI machine named DABUS (Device for the Autonomous Bootstrapping of Unified Sentience), created by Dr. Thaler.
DABUS autonomously generated two novel products: a fractal drink container entitled ‘Food Container’ and a flashing beacon called ‘Devices and Methods for Attracting Enhanced Attention’. Dr. Thaler applied for patents in the UK Intellectual Property Office (UKIPO), naming DABUS as the inventor. However, the applications were withdrawn on the grounds of non-compliance, as the law required the inventor to be a “natural person”.
Dr. Thaler appealed to the High Court and subsequently to the Court of Appeal, both of which dismissed his case. The courts held that under the Patents Act 1977, only natural persons could be inventors, although both inventions themselves were technically patentable under s.1 of the Act.
The case reached the UK Supreme Court in December 2023. The Court dismissed the appeal, holding that:
- The term ‘inventor’ in the Patents Act 1977 refers to a natural person and cannot extend to autonomous machines.
- Only a ‘natural person’ can devise an invention, and this interpretation was the only reasonable one under the Act.
- A machine, such as DABUS, could not be an ‘inventor’.
- Recognising DABUS an inventor would be legally flawed and an impossible proposition.
The Supreme Court also noted that Parliament had not contemplated the possibility of a machine acting autonomously or being recognised as an inventor when drafting the Act.
Globally, the debate has continued.
Applications naming DABUS as an inventor have been rejected by the United States Patent and Trademarks Office (USPTO), the European Patent Office (EPO), and authorities in Germany, Taiwan, and New Zealand, all citing the requirement that inventorship demands “mental conception”, a capacity unique to the human mind. However, South Africa became the first country to grant a patent to DABUS as an inventor — a decision that sparked widespread debate.
The DABUS case highlights the broader challenge: as AI becomes increasingly capable of automating complex tasks—ranging from drug discovery and engineering to creating art and music—should it be recognised as more than a tool? If AI systems generate other AI systems or create patentable products, should they be considered inventors, or should inventorship always rest with a natural person?
Jurisprudence currently recognises two categories: a natural person (a living human being) and a legal person (entities such as companies). If companies can be recognised, some argue it may be time to debate whether AI could one day be accorded a similar status.
As laws are typically prospective in nature, legal frameworks must evolve to keep pace with the rapid growth of the global AI industry. Studies indicate that deep learning models such as generative AI, capable of generating high-quality text, images, and other content, are now widely accessible. This development introduces complex legal dilemmas around copyright infringement, as AI-generated content frequently draws upon existing material without proper attribution or authorisation from the original creators.
The absence of a clear legislation could lead to further conflicts over ownership of intellectual property. Robust governance mechanisms and regulations for AI are urgently needed to address questions of authorship, inventorship, and accountability.
In India, the National Strategy for AI, popularly known as #AIforAll, issued by Niti Aayog, remains the primary guiding framework for the country’s AI ecosystem. The Parliamentary Standing Committee in its Report No. 161 recognised the potential of AI and machine learning (ML) to drive economic growth and recommended extending IP protections to AI-based inventions.
However, current laws—the Patents Act 1970 and the Copyright Act 1957—are not equipped to facilitate inventorship, authorship, or ownership by AI. For example, Section 2(d)(vi) of the Copyright Act designates the “person who causes the creation” of a computer-generated work as the author, reaffirming human agency.
Regarding inventions, Section 3(k) of the Patents Act 1970 excludes mathematical methods, computer programmes per se, and algorithms from patentability. In early 2025, the Office of the Controller General of Patents, Designs, and Trademarks (CGPDTM) released Draft Guidelines for Examination of Computer-Related Inventions (CRI) for public consultation, aiming to clarify the patentability of such inventions.
The pressing question remains: if AI continues to generate novel and patentable products, who can truly be recognised as the ‘inventor’?
The author is a Teaching Associate in Organisation and Leadership Studies at S.P. Jain Institute of Management & Research (SPJIMR). Views are personal.
