Author: Chhavi Das, ILS Law College
To the Point
At the heart of patent law lies a surprisingly human-centric principle: inventors must be people. As AI grows increasingly capable of generating novel inventions, the legal world faces a serious dilemma, should machines be granted inventorship? While jurisdictions like the US, UK, EU, India, and Japan have doubled down on the human-only inventor rule, voices are rising that call for reform. They argue that ignoring AI-generated inventions might hold back innovation. However, current laws are rooted in the assumption that invention is a human endeavor, and any shift will likely need legislative, not judicial, action.
Use of Legal Jargons
Within the law of patents, inventorship is the legal acknowledgment of the person who originates the first and new idea of an invention. This is a basic term, since only inventors or persons to whom the rights are transferred legally can be granted patent protection. The majority of legal systems require that the inventor be a natural person, i.e., a human, and not an artificial person or AI program. This is closely related to the idea of legal personality, the legal standing that is necessary in order to own property, sign letters, transfer rights, or be a party to court proceedings, all things which AI does not yet possess.
Another crucial term is assignee, meaning someone or something that acquires rights to a patent from the initial inventor, either by contract or employment. Still, the assignee will have to trace back to an individual human inventor. Courts most frequently resort to statutory interpretation in order to shed light on ambiguous words in law. In such arguments, they’ve consistently interpreted the term “individual” or “person” in patent laws to be an individual human being.
The term true and first inventor occurs under Indian law in Section 2(1)(y) of the Patents Act, 1970, and refers to those who did not contribute intellectually to the conception of the invention, be it financiers, corporations, or machinery. Concepts such as substantive examination refer to the mechanism by which patent offices determine if an invention fulfills aspects such as novelty, usefulness, and lack of obviousness. South Africa, on the other hand, follows a system of deposits, or rather, it records patents without thoroughly scrutinizing the validity of inventorship, wherein DABUS was able to be credited as inventor there, albeit without precedent in the law or judicial review.
The Proof
Legal Texts: Patent statutes generally do not explicitly define “inventor,” but use terms like “individual” or “person” in context. For instance, 35 U.S.C. §115 (not cited above) and related rules require each inventor to submit an oath under penalty of perjury. The UK Patents Act 1977 requires an application to name the inventor and states failure to name a true inventor renders the application void. India’s Patents Act, 1970 similarly limits applications to the true and first inventor or their assignee (Section 6). Section 2(1)(y) defines “true and first inventor” by excluding others (e.g. financial contributors), reinforcing that inventor refers to someone who actually made the invention. By contrast, none of these provisions contemplates a machine. (India’s definition of “person” includes corporations and governments, but even that does not suggest allowing non-human inventors. In short, the statutory frameworks presuppose human innovators.
Court and Office Decisions: Courts have consistently interpreted these statutes to exclude AI. In Thaler v. Vidal (Fed. Cir. 2022), the U.S. Court of Appeals examined the Patent Act’s language and concluded “there is no ambiguity: the Patent Act requires that inventors must be natural persons; that is, human beings”. Judge Stark noted the Act uses pronouns “himself” or “herself” (never “itself”) when referring to an inventor. The Board of Appeal of the European Patent Office reached the same conclusion in 2021: designating an AI as inventor “did not meet the requirements” of EPC Rule 19, because an inventor must be a human being and a machine has no legal personality to transfer rights.
Patent offices have put these rulings into practice. The U.S. Patent Office and courts rejected Dr. Thaler’s DABUS applications, as did the Virginia federal court (Thaler v. Iancu, 2021). In the UK, the Intellectual Property Office and then the UK courts followed suit, culminating in the Supreme Court’s 2023 ruling. Australia’s Patent Office and courts likewise require a human inventor. In 2024 Tokyo, a District Court explicitly interpreted Japan’s Patent Act to mean “an inventor… should be a natural person”.
Precedents: In the absence of direct AI cases, courts have long held only humans can qualify as inventors. For example, the Indian IPAB has ruled that neither a mere financier nor a corporation can be named sole inventor. These precedents were cited in objections to AI inventor claims in India. To date, no judicial precedent has recognized a machine as inventor, aside from a short-lived Australian decision (Beach J, 2021) that was later overturned.
Inventorship vs. Ownership: Courts also consider who can own an AI-generated invention. In the UK case, for example, it was held that owning DABUS did not entitle Thaler to the patents, since no statute allows a machine’s owner to claim its inventions. Similarly, Indian law’s Sections 25 and 83 tie patent rights to named inventors and assignees, bolstering the view that a human must stand in place of the machine.
Taken together, the “proof” from statutes and decisions is that current patent systems expect a human inventor. As one legal analysis concludes, “as it stands now, AI is not recognized as an inventor under the patent laws of the United States, Europe, and Australia”. Any change to this rule, most acknowledge, would have to come from legislation or treaty revisions.
Abstract
With AI systems like DABUS generating inventions autonomously, the debate over whether machines can be named inventors has intensified. Almost all major jurisdictions maintain that only humans can hold this title. This article explores why courts and patent offices are reluctant to change, what the legal framework currently says, and how lawmakers might respond in the future. While the South African Patent Office stands as an exception, the overwhelming global view is that patent systems remain human-centered.
Case Laws (Major Decisions)
India
DABUS Patent Application (2020, Pending) – A patent application was made with DABUS as the inventor. The Indian Patent Office has not yet granted the patent, and objections have been filed on the basis that Sections 2(1)(y) and 6 of the Patents Act, 1970 confine the status of an inventor exclusively to a natural person. The case is pending, but the legislative framework strongly prejudices against AI inventorship.
R.A. Mashelkar Committee Reports – Although not a case, policy reviews by the Indian government (particularly the 2004 and 2018 reports) have highlighted human contribution as the core of patent law. They have impacted legislative interpretation in rejecting non-human inventorship.
Dr. Kankanala’s Opposition (2023) – IP expert Dr. Kalyan Kankanala raised a formal objection in the DABUS application on the grounds that the actual and first inventor needs to prove mental application, which cannot be achieved with AI systems under Indian law.
USA:
In Thaler v. Vidal (2022), the U.S. Federal Circuit held that “individual” in the Patent Act means only human beings, confirming that AI systems such as DABUS cannot be inventors. Previous rulings by a Virginia federal court also dismissed AI inventorship.
United Kingdom:
In its 2023 judgment in Thaler v. Comptroller-General of Patents, the UK Supreme Court affirmed that, under the Patents Act, inventorship is limited to natural persons. The Court emphasised that any change to the existing legal framework must come through legislative action by Parliament.
Australia:
The Full Federal Court in Commissioner of Patents v. Thaler (2022) held that an inventor under the Patents Act 1990 must be a human being. The Court emphasized patent law to presume inventions originate from human minds and that there was no legislative intention to consider AI.
European Patent Office (EPO):
In Decision G 0001/22 (J 0008/20), the Legal Board of Appeal reiterated that in the European Patent Convention, an inventor has to be an individual. DABUS applications were rejected since a machine cannot meet procedural or legal duties of inventorship.
Japan:
In May 2024, the Tokyo District Court ruled that inventorship under Japan’s Patent Act is understood as a product of human creative endeavor, drawing on the Intellectual Property Basic Act. The court did not accept AI as an inventor and left any policy adjustment to the legislature.
Conclusion
The bottom line is that, for now, patent laws assume inventors are people. Courts and patent offices in the US, UK, Europe, Australia, Japan, China, and elsewhere have all required identifying a natural person on the patent application.
AI systems remain tools, not legal “persons” capable of owning IP. Whether this strict human-centric view should change is a policy question. For example, the UK Supreme Court explicitly noted it was not its role to rewrite patent law for the AI and a Japanese court similarly said any change should occur through democratic legislation.
This evolving issue remains under active observation by legislative bodies and patent authorities globally. Some have suggested amending laws to allow human patenting of fully autonomous AI inventions – perhaps by defining a new category of inventor or by requiring at least “significant” human contribution. Others propose leaving the law as is and using existing doctrines: for instance, if a human designs or trains the AI, that human is the true inventor.
In the meantime, inventors and companies using AI simply name a human developer, operator, or owner on the patent applicationreedsmith.com. In summary, the debate is not settled. The DABUS case has highlighted the gap between rapidly advancing machine creativity and static legal categories. That said, most courts and patent offices agree: as things stand, AI just doesn’t qualify as an inventor. Future changes, if any, will require explicit legal reform. Scholars advise that we balance encouragement of innovation with the fundamental patent bargain (public disclosure for limited monopoly). Until then, humans will remain in the inventor’s seat.
FAQS
• Has any nation recognized AI as an inventor?
South Africa is currently the sole jurisdiction where a patent has been granted with an artificial intelligence system, DABUS, designated as the inventor. All the important jurisdictions – the US, UK, EU, Australia, Japan, China, and India – have declined to do so on the grounds that only humans can be inventors.
• What is the DABUS case?
DABUS is an AI created by Dr. Stephen Thaler, credited with generating inventions autonomously. Thaler filed patent applications worldwide naming DABUS as the inventor. Except for South Africa, courts and patent offices globally have rejected the idea of AI inventorship.
• How does Indian law treat AI inventors?
Indian legislation does not specifically refer to AI, but the Patents Act insists on a “true and first inventor” being human. The use of DABUS in India was opposed on such grounds, and the case remains pending.
• To whom should the patent go if the invention has been made by an AI?
Typically, a human who participated in making or operating the AI—such as a developer or operator—needs to be credited as inventor. Courts won’t accept shifting rights from AI to its owner without statutory backing.
• Will patent law be modified to enable AI inventors?
Perhaps in the future, but not currently. Courts indicate that this is something lawmakers need to decide. Some nations like China are discussing guidelines, but as yet no great nation has changed its laws to recognize AI inventors.
