Skip to content.

Do generative AI inventions and works qualify for patents and copyrights? The Thaler and SURYAST decisions

Artificial intelligence (AI) systems and in particular generative AI (GenAI) systems have raised the question as to whether technical advances in the useful arts or synthetic content generated using these tools can qualify for patent or copyright protection. Recent decisions in both the patent and copyright fields have denied protection for otherwise patentable inventions and copyright works where the sole claimed inventor or author is identified as an artificial intelligence system. The recent UK Supreme Court decisions in Thaler (Appellant) v Comptroller-General of Patents, Designs and Trade Marks [2023]UKSC 49 and the US Copyright Review Board in Refusal to Register SURYAST are illustrative.

Do generative AI inventions qualify for patent protection

Prior to the UK Supreme Court decision in Thaler, courts that have considered the question have largely held that an inventor of otherwise patentable subject matter must be a natural person. This was the decision reached in the United States by the Federal Circuit Court of Appeals in Thaler v Vidal 43 F.4th 1207 (2022) which ruled that Section 271 of the Patent Act requires that an inventor be a human being. The European Patent Office reached the same conclusion in J 0008/20 (Designation of inventor/DABUS) of 21.12.2021 as did Australia’s Full Court in Commissioner of Patents v Thaler ‑ [2022] FCAFC 62.

The UK Court of Appeal in Thaler v Comptroller General of Patents Trade Marks And Designs [2021] EWCA Civ 1374 (21 September 2021), the decision which was appealed to the UK Supreme Court, also ruled that an inventor must be a human being.

Unsurprisingly, the UKSC concluded that, as a matter of statutory construction, the meaning of “inventor” under the UK Patents Act 1977 had to be a natural person.

I should say straight away that this issue has been decided against Dr Thaler at every level in these proceedings and by every judge who has considered it. Dr Thaler has argued throughout that the technical advances and the new products described and disclosed in the applications were devised by DABUS, and that DABUS was their inventor. As I have indicated, the Comptroller accepts for the purposes of these proceedings the substance of the factual case advanced by Dr Thaler, namely that DABUS created or generated the technical advances described and disclosed in the applications and did so autonomously using AI, but the Comptroller has never accepted (and disputes any suggestion) that this renders DABUS an inventor within the meaning of the 1977 Act.

In my judgment, the position taken by the Comptroller on this issue is entirely correct. The structure and content of sections 7 and 13 of the Act, on their own and in the context of the Act as a whole, permit only one interpretation: an inventor within the meaning of the 1977 Act must be a natural person, and DABUS is not a person at all, let alone a natural person: it is a machine and on the factual assumption underpinning these proceedings, created or generated the technical advances disclosed in the applications on its own. Here I use the term “technical advance” rather than “invention”, and the terms “create” or “generate” rather than “devise” or “invent” deliberately to avoid prejudging the first issue we have to decide. But it is indisputable that DABUS is a machine, not a person (whether natural or legal), and I do not understand Dr Thaler to suggest otherwise…

In all these circumstances the Comptroller was right to decide that DABUS is not and was not an inventor of any new product or process described in the patent applications. It is not a person, let alone a natural person and it did not devise any relevant invention. Accordingly, it is not and never was an “inventor” for the purposes of section 7 or 13 of the 1977 Act.

Dr Thaler had also argued that he was nevertheless the owner of the “invention” under the “doctrine of accession”. This argument had been rejected by Arnold JA in the Court of Appeal and it was also rejected by the UKSC.[i]

Commentary re computer generated patentable inventions

In Thaler, the UKSC made it clear that its decision was not dealing with the larger policy question of whether, or in what circumstances, an AI generated non-obvious technical advance should be patentable. The reasons for decision were limited to the much more focused question of the correct interpretation and application of the relevant provisions of the 1977 Act to the applications made by Dr Thaler.

The Comptroller has emphasised, correctly in my view, that this appeal is not concerned with the broader question whether technical advances generated by machines acting autonomously and powered by AI should be patentable. Nor is it concerned with the question whether the meaning of the term “inventor” ought to be expanded, so far as necessary, to include machines powered by AI which generate new and non-obvious products and processes which may be thought to offer benefits over products and processes which are already known.

These questions raise policy issues about the purpose of a patent system, the need to incentivise technical innovation and the provision of an appropriate monopoly in return for the making available to the public of new and non-obvious technical advances, and an explanation of how to put them into practice across the range of the monopoly sought. It may be thought that the rapid advances in AI technology in recent times render these questions even more important than they were when these applications were made.

The UKSC also noted that the decision did not address the very practical, important and different question as to whether, or in what circumstances, a natural person could obtain a patent when a technical advance is made using a generative AI system:

It follows but is important to reiterate nonetheless that, in this jurisdiction, it is not and has never been Dr Thaler’s case that he was the inventor and used DABUS as a highly sophisticated tool. Had he done so, the outcome of these proceedings might well have been different.

Similar questions were raised by the Full Court of Australia in Commissioner of Patents v Thaler.

In our view, there are many propositions that arise for consideration in the context of artificial intelligence and inventions. They include whether, as a matter of policy, a person who is an inventor should be redefined to include an artificial intelligence. If so, to whom should a patent be granted in respect of its output? The options include one or more of: the owner of the machine upon which the artificial intelligence software runs, the developer of the artificial intelligence software, the owner of the copyright in its source code, the person who inputs the data used by the artificial intelligence to develop its output, and no doubt others. If an artificial intelligence is capable of being recognised as an inventor, should the standard of inventive step be recalibrated such that it is no longer judged by reference to the knowledge and thought processes of the hypothetical uninventive skilled worker in the field? If so, how? What continuing role might the ground of revocation for false suggestion or misrepresentation have, in circumstances where the inventor is a machine?

Those questions and many more require consideration. Having regard to the agreed facts in the present case, it would appear that this should be attended to with some urgency…

However, the characterisation of a person as an inventor is a question of law. The question of whether the application the subject of this appeal has a human inventor has not been explored in this litigation and remains undecided. Had this question been explored, it may have been necessary to consider what significance should be attributed to various matters including the (agreed) facts that Dr Thaler is the owner of the copyright in the DABUS source code and the computer on which DABUS operates, and that he is also responsible for the maintenance and running costs.

As noted above, the Court of Appeal and the UKSC in Thaler ruled that a person that uses a GenAI system cannot claim any patent rights to an invention under the doctrine of accession. Neither decision, however,  canvassed whether under the doctrine of accession a tangible product produced by a GenAI system might be subject to some other property right or who would own the right. Neither did they consider whether that right could be violated if the product was dealt with without authorization under a tort conversion theory, or whether any such right would be pre-empted by an IP right.

You can watch videos of the argument before the UKSC here.

It has been a long standing tenet of copyright law that an original work must be created by an author. It is of course clear that a person can be an author even if a computer based tool is used to assist in creating the work as long as the person controlling the computer is seen as directing or fashioning the material form of the work. But, if the person does not contribute sufficient independent intellectual effort to satisfy the Act’s requirement for originality, the individual using the computer would likely not be considered its author. This is the basis for the long understood distinction between computer assisted and computer generated works.[ii]

In the US there have been several decisions from the Copyright Office refusing to recognize copyrights for computer generated works, but allowing copyrights as long as the AI generated portions are disclaimed. The Copyright Office position was followed by a District Court in the Thaler v Perlmutter case.

The US Copyright Office Review Board confirmed the position of the US Copyright Office in a recent case refusing to register the 2 dimensional artwork called SURYAST produced by Mr Sahni.

The two dimensional artwork was produced by providing a generative AI tool called RAGHAV with a base image (Mr. Sahni’s original photograph), a style image (Vincent van Gogh’s The Starry Night), and an undisclosed numerical value for the strength of the style transfer. RAGHAV then generated SURYAST work using the GenAI system RAGHAV.

The original photograph used as the base work, the van Gogh style image, and the work submitted for registration are shown below.

Starry Night
Suryast (the Work)

The Copyright Review Board refused to issue a copyright registration on the basis that there was insufficient human authorship in the creation of SURYAST.

After considering the information provided by Mr. Sahni regarding his creation of the Work, including his description of RAGHAV, the Board concludes that the Work is not the product of human authorship. Specifically, the Board finds that the expressive elements of pictorial authorship were not provided by Mr. Sahni. As Mr. Sahni admits, he provided three inputs to RAGHAV: a base image, a style image, and a “variable value determining the amount of style transfer.” Because Mr. Sahni only provided these three inputs to RAHGAV, the RAGHAV app, not Mr. Sahni, was responsible for determining how to interpolate the base and style images in accordance with the style transfer value. The fact that the Work contains sunset, clouds, and a building are the result of using an AI tool that “generate[s] an image with the same ‘content’ as a base image, but with the ‘style” of [a] chosen picture.” But Mr. Sahni did not control where those elements would be placed, whether they would appear in the output, and what colors would be applied to them.

Mr Sahni argued that RAGHAV was merely a computer assisted tool, the use of which should qualify the output for copyright protection. The Board did not accept this argument.

The Board is not convinced by Mr. Sahni’s description of RAGHAV as “an assistive tool” that works similarly to “a camera, digital tablet, or a photo-editing software program.” In his Second Request, Mr. Sahni now describes RAGHAV as merely “mechanically appl[ying] the colors, shapes, and style as directed, which is not any different from, for example, Adobe Photoshop applying red and blue shades to a photograph based on a user’s command.” This description inaccurately minimizes RAGHAV’s role in the creation of the Work and conflicts with other information in the record. As Mr. Sahni stated in his initial explanation, RAGHAV operates by “generat[ing]” a new pictorial image based on features learned from user-provided images. The underlying research that RAGHAV was built on is premised on the same functionality: it is the AI model, not its user, that “predict[s] stylizations for paintings and textures never previously observed,” and that predictive function is tied to “the proximity of the [style image] to styles trained on by the model.”

Here, RAGHAV’s interpretation of Mr. Sahni’s photograph in the style of another painting is a function of how the model works and the images on which it was trained on—not specific contributions or instructions received from Mr. Sahni. While Mr. Sahni selected the numerical variable for the “strength” of the style, that choice alone is insufficient to warrant copyright protection. As noted above, selecting a single number for a style filter is the kind of de minimis authorship not protected by copyright.

While Mr Sahni was unable to establish that he had contributed sufficient authorship in this instance, as Edward Lee argued in an article “Prompting Progress: Authorship in the Age of AI”, there will likely be cases where a person should be able to show a sufficient degree of control over what is ultimately created using a GenAI system to give rise to a protectable copyright right. This would be consistent with a recent case decided by a Chinese court.


[i] According to the UKSC:

The DABUS inventions are, he says, the fruits of (in the sense they were produced by) the DABUS machine that he owns and further, that DABUS was designed to make inventions and so these fruits were by no means unexpected. He also contends that he was and remains the first person to possess the inventions and this provides a proper basis for their ownership. In short, he contends that he derived title by operation of a rule of law (the doctrine of accession) that satisfied the terms of section 7(2)(b) of the Act and conferred on him the right to apply for and secure the grant of patent protection for any inventions made by DABUS.

In my view there are two fundamental problems with these submissions. The first is that they assume that DABUS can itself be an inventor within the meaning of the 1977 Act. But that assumption is not correct for the reasons I have given. DABUS is a machine and not a person. That was reason enough for the Hearing Officer for the Comptroller to reach the conclusion he did. Indeed, it was itself fatal to the applications. There was no inventor through whom Dr Thaler could claim the right to obtain a patent for any technical advance described in those applications.\

The second is that it mischaracterises an invention as being or amounting to tangible property such that title to it can pass, as a matter of law, to the owner of the machine which, on this assumption, generated it. I accept, of course, that the 1977 Act refers at times to the property in an invention. As we have seen, it does so in, most importantly, section 7(2)(b), and this is the provision on which Dr Thaler places particular reliance. The 1977 Act also contemplates in, for example, section 39, that an invention may be taken as “belonging” to a person, such as an employee or an employer.

One must be careful to understand what this means, however. The right we are concerned with, as conferred by the 1977 Act, is a right to apply for a patent for what is said to be an invention and, if it is patentable and satisfies the other requirements of the Act, to secure the grant of a patent on that application. But I am satisfied that Dr Thaler has not identified any basis in law on which he acquired such a right through his ownership of DABUS.

In particular, Dr Thaler’s reliance on the doctrine of accession in this context is misguided. The doctrine concerns new tangible property produced by existing tangible property. Dr Thaler contends that, upon the application of this doctrine, the owner of the existing property also owns the new property. In this way, the farmer owns the cow and the calf. By analogy, Dr Thaler continues, he, as owner of DABUS, is the owner of all rights in all developments made by DABUS.

We are not concerned here with a new item of tangible property produced by an existing item of tangible property, however. We are concerned with what appear (and which for present purposes we must assume) to be concepts for new and non-obvious devices and methods, and descriptions of ways to put them to into practice, all of which, so Dr Thaler maintains, have been generated autonomously by DABUS. There is no principled basis for applying the doctrine of accession in these circumstances.

[ii] See for example these Commonwealth cases, Harmony Consulting Ltd. v. G.A. Foss Transport Ltd., 2012 FCA 226; Telstra Corporation Limited v. Phone Directories Company Pty Ltd., [2010] FCAFC 149 (15 December 2010); Acohs Pty Ltd v. Ucorp Pty Ltd., [2010] FCA 577 (10 June 2010); IceTV Pty Ltd. v. Nine Network Australia Pty Ltd., [2009] HCA 14, Asia Pacific Publishing Pte Ltd v Pioneers & Leaders (Publishers) Pte Ltd [2011] SGCA 37, Camlin Pvt. Ltd. vs National Pencil Industries on 7 November, 1985, AIR 1986 Delhi 444.


This article was first posted on



Stay Connected

Get the latest posts from this blog

Please enter a valid email address