Abstract

Artificial intelligence can generate technical designs, identify chemical compounds and propose solutions that may not have been anticipated by human researchers. This raises a difficult question for patent law: can an AI system be legally named as an inventor? This article examines Indian patent law and comparative decisions from the United Kingdom, United States, Australia, Europe, Germany, New Zealand and Switzerland. It distinguishes the technical ability to generate an invention from legal recognition as an inventor, and considers the implications for entitlement, human contribution and patentability.

1. Introduction

Patent law traditionally assumes that an invention can be attributed to a human inventor. In exchange for public disclosure, patent protection grants exclusive rights for a limited period. AI systems can analyse large datasets, identify patterns and generate technical solutions, sometimes with limited human intervention. If an AI system produces a useful new material or design, the question is not only whether the result is new and non-obvious, but who can legally be named as inventor and who may own the resulting rights.

2. AI-Assisted and AI-Generated Inventions

An AI-assisted invention involves a human who uses AI as a tool but makes a meaningful contribution to the claimed inventive concept. An AI-generated invention raises a harder issue where the inventive concept is said to have been produced autonomously and no human can properly be identified as its deviser. AI use alone does not make an invention unpatentable; inventorship, patentability and ownership are separate questions.

Traditional inventorship decisions help analyse human contribution even though they predate generative AI. In Burroughs Wellcome Co. v Barr Laboratories, Inc., 40 F.3d 1223 (Fed. Cir. 1994), the court treated conception as central to inventorship: a person must form a definite and permanent idea of the complete invention, rather than merely identify a research goal. In Pannu v Iolab Corp., 155 F.3d 1344 (Fed. Cir. 1998), the court explained that a joint inventor must make a significant contribution to conception and do more than explain well-known concepts. Applied cautiously, these principles can help distinguish a human who meaningfully devises a claimed AI-assisted invention from a person who merely asks a system a broad question.

3. The Indian Legal Framework

The Patents Act, 1970 governs patent protection in India. Section 2(1)(j) defines an invention by reference to a new product or process involving an inventive step and capable of industrial application. Section 2(1)(ja) addresses inventive step, including technical advancement or economic significance and non-obviousness to a person skilled in the art. Section 6 identifies persons entitled to apply, including the true and first inventor, an assignee and, in specified circumstances, a legal representative.

The Act distinguishes inventorship from ownership. A company may own a patent through a valid assignment, but this does not make the company the inventor.

Section 3(k), which excludes mathematical methods, business methods, computer programmes per se and algorithms, is a separate issue. An AI-related invention must be assessed on its claimed subject matter and technical implementation; AI involvement alone does not decide patentability.

The Parliamentary Standing Committee on Commerce discussed AI and intellectual property in its 161st Report, Review of the Intellectual Property Rights

Regime in India (2021). Its observations contribute to policy debate, but committee recommendations are not themselves enacted law and do not confer inventorship rights on AI systems.

4. The DABUS Litigation

The international debate was sharpened by applications filed by Dr Stephen Thaler naming DABUS—Device for the Autonomous Bootstrapping of Unified

Sentience—as inventor of inventions including a food container and a flashing light. Thaler argued that DABUS had generated the inventions and that his ownership of the system entitled him to the patents.

In Thaler v Comptroller-General of Patents, Designs and Trade Marks [2023] UKSC 49, the UK Supreme Court held that an inventor under the Patents Act 1977 must be a natural person. It also rejected the argument that ownership of DABUS automatically gave Thaler entitlement to patents for inventions attributed to it.

The Court interpreted the legislation; it did not decide that AI is incapable of producing technically inventive results.

In the United States, Thaler v Vidal, 43 F.4th 1207 (Fed. Cir. 2022), held that the Patent Act’s reference to an individual required a natural person. The court did

not decide the philosophical question of AI consciousness; it held that the statute requires a human inventor and that the applications named none. The US Supreme Court later declined review.

The Australian litigation shows why appellate history matters. In Thaler v Commissioner of Patents [2021] FCA 879, a single judge initially held that an AI system could be an inventor. That ruling was overturned by the Full Court in Commissioner of Patents v Thaler [2022] FCAFC 62, which held that the statutory

framework did not permit a machine to be named inventor. The appellate ruling is the operative Australian position.

The European Patent Office Legal Board of Appeal reached a similar result in J 8/20 and J 9/20. It held that the European Patent Convention requires an inventor

to be a person with legal capacity and rejected the argument that ownership of DABUS made Thaler its successor in title. The decisions addressed designation and entitlement, not a blanket prohibition on inventions developed with AI assistance.

In Thaler v Commissioner of Patents [2023] NZHC 554, the New Zealand High Court held that the statutory meaning of actual deviser did not include an AI system. The Court considered that a change allowing AI inventorship should be made by Parliament rather than through judicial reinterpretation.

Germany’s Federal Court of Justice, decision X ZB 5/22 (11 June 2024), likewise held that only a natural person may be named as inventor. It nevertheless recognised that a human who used AI to find the technical teaching may be named if the human contribution is properly described. This illustrates the continuing distinction between AI as a tool and AI as the named inventor.

In Switzerland, Federal Administrative Court decision B-2532/2024 (26 June 2025) rejected naming DABUS as inventor and rejected leaving the inventor field blank, while allowing an alternative request to name Thaler. The court’s approach focused on whether a human substantially contributed to the AI data-treatment process and recognised the output as an invention. It did not recognise AI itself as inventor.

5. Human Contribution: Further Patent-Law Principles

Several established US cases help explain why the precise human contribution matters. In Fina Oil & Chemical Co. v Ewen, 123 F.3d 1466 (Fed. Cir. 1997), the court explained that helping reduce an idea to practice does not by itself make someone a joint inventor; the contribution must relate to conception. In Eli Lilly &

Co. v Aradigm Corp., 376 F.3d 1352 (Fed. Cir. 2004), the court’s analysis illustrates the importance of evidence linking a person’s contribution to the claimed invention rather than merely to the wider research project.

In Ethicon, Inc. v United States Surgical Corp., 135 F.3d 1456 (Fed. Cir. 1998), the court addressed joint inventorship and explained that a qualifying co-inventor

need not contribute to every claim. This is relevant where several researchers work with AI at different stages: each person’s contribution must be assessed againstthe invention, not their job title or seniority.

In Yeda Research & Development Co. Ltd. v Rhône-Poulenc Rorer International Holdings Inc. [2007] UKHL 43, the House of Lords considered entitlement and the identity of the person who actually devised the invention. The case reinforces that patent entitlement depends on the statutory scheme and the facts of inventorship, rather than employment or funding relationships alone.

In University of Utah v Max-Planck-Gesellschaft zur Förderung der Wissenschaften e.V., 734 F.3d 1315 (Fed. Cir. 2013), the Federal Circuit considered inventorship, assignments and ownership in a research dispute. It illustrates that ownership must rest on a legally sufficient route to title. Similarly, Board of

Education v American Bioscience, Inc., 333 F.3d 1330 (Fed. Cir. 2003), underscores that patent title cannot be inferred merely from an organisation’s involvement in research.

6. Implications for India and Possible Reform

The comparative cases generally reject AI itself as a named inventor under current statutory language, while leaving open the possibility of patent protection for qualifying inventions devised by humans using AI. The decisions differ in reasoning and procedural posture, so they should not be treated as identical or as binding Indian precedent. The Indian Patents Act and Indian judicial decisions remain controlling in India.

Research organisations should preserve records of prompts, model outputs, experiments, human decisions and modifications to the claimed technical features. Such records may help establish who conceived the invention, but documentation cannot create a legal entitlement where the statute does not recognise one.

Decisions such as Pannu, Burroughs Wellcome and Fina Oil can offer useful analytical principles for assessing human contribution, although they arose in different factual and legal settings.

Supporters of reform argue that uncertainty may discourage investment in AI-driven research. Critics emphasise accountability, ownership uncertainty and the risk of large volumes of AI-generated applications. India could clarify inventorship and entitlement rules while preserving disclosure requirements and the separate tests of novelty, inventive step and industrial applicability. Reform need not grant AI legal personality; it could instead clarify when human contributions support a patent application.

7. Conclusion

The DABUS litigation demonstrates that legal inventorship is not identical to technical generation. The UK, US, Australian, EPO, New Zealand and German authorities do not recognise AI itself as inventor under their current frameworks, while the Swiss decision illustrates a route based on identifying a meaningful human contribution. Traditional inventorship cases also show that a person must contribute to the conception of the claimed invention; funding, routine testing or ownership alone may not suffice.

For India, the central challenge is to allocate rights fairly without weakening disclosure, accountability or public access to knowledge. AI-assisted inventions should be assessed on their facts and claims, while any future reform should provide clear rules for human contribution and entitlement. The goal should be a patent system that encourages genuine innovation without granting exclusive rights on an uncertain legal basis.