Copyright, AI and generative Art

04.10.23

‘What are the key elements of creative work? Where is the boundary between human creativity and automated work? What is the difference between AI work and AI art?

Peter Howitt

Managing Director

TL;DR

This article looks at the current state of copyright law in the US and UK for AI art and generative creative work. It is written by a lawyer and creative user of AI tools – it should be read by artists, media and tech companies and lawyers.

If are busy right now (or have a short attention span :)) you can watch the brief video I made:

Thanks for your time 🙂

If you have any issues with playback, the video is also available on YouTube.

US court and Copyright Office considers AI works and Copyright

“Gibraltar City Walls”, Peter Pink-Howitt

Recently a US case (Thaler v Comptroller-General of Patents, Designs and Trademarks), considered the ability of algorithmically generated artwork to attract copyright protection.

The court found against the appellant based on a principle of human authorship that – though not enshrined in copyright law – is a working principle used by the US Copyright Office.

The case reaffirmed US guidance issued in March 2023 as part of the US Copyright Office’s AI initiative – that copyright only protects material that is the product of human creativity. The Judge found that the concept of authorship has been “synonymous with human creation over time” and “[human] authorship is a bedrock requirement of copyright“.

The US district court stated that:

“[c]opyright has never stretched so far, however, as to protect works generated by new forms of technology operating absent any guiding human hand.”

(Note: That is not the case with copyright laws in some other countries.)

In another Copyright Office ruling, an award-winning generative artwork Théâtre D’opéra Spatial by Jason Matthew Allen was denied copyright.

“The Board finds that the Work contains more than a de minimis amount of content generated by artificial intelligence (“AI”), and this content must therefore be disclaimed in an application for registration. Because Mr. Allen is unwilling to disclaim the AI-generated material, the Work cannot be registered as submitted.”
US Copyright Office Ruling

Mr Allen insists he will fight on:

“Allen was dogged in his attempt to register his work. He sent a written explanation to the Copyright Office detailing how much he’d done to manipulate what Midjourney conjured, as well as how much he fiddled with the raw image, using Adobe Photoshop to fix flaws and Gigapixel AI to increase the size and resolution. He specified that creating the painting had required at least 624 text prompts and input revisions.”
WIRED – Why This Award-Winning Piece of AI Art Can’t Be Copyrighted

Théâtre_D’opéra_Spatial – Colorado State Fair, Matthew Allen

In my opinion, the advance of AI and the IP Devil risk (see “Why should we limit the scope of copyright to works with sufficient human input and effort?” below) means an examination of the extent of human input in AI-created works should always be a requirement in an IP assessment for creative work.

That said, in the case of Matthew Allen and his iterated creation, the Copyright Office is fundamentally wrong in its understanding of technology, the creative process and, with respect, the law. Mr Allen can demonstrate sufficient effort, expertise and creativity in how he used Midjourney to create his work. I expect a US court to make this clear when the matter is heard. The current approach to AI art by the Copyright Office will have the effect of chilling human creativity using AI tools.

The decision is also inconsistent with the Office’s guidance in other areas, e.g., photography:

“A work is original if it is independently created and is sufficiently creative. Creativity in photography can be found in a variety of ways and reflect the photographer’s artistic choices like the angle and position of subject(s) in the photograph, lighting, and timing.”
US Copyright Office

Copyright law does not suggest that the success of your claim for a photograph depends on whether you use a 35mm or old-school SLR versus a digital camera or smartphone (with many more advanced settings than were available in older cameras). Nor does it suggest that copyright will be determined by whether you removed any advanced technology features or on how many of the settings you changed.

The Copyright Office is not taking a technology-neutral approach to the use of AI (or algorithmic) tools. For other creative expressions (music, photography, writing) the test is effectively whether there is de minimis human creativity. For works using AI tools, the Copyright Office has made the test a different one i.e. whether there is no more than de minimis technological involvement.

If Mr Allen persists with his battle for legal recognition, we expect soon to have US legal precedent and guidance on what is sufficient human involvement in an AI process for work to be considered as ‘human authored’. Given the risks of AI, the Copyright Office would no doubt welcome some guidance too.

In fairness, the Copyright Office recognises the need for dialogue about the impact of AI tools on the existing regime. It recently published a Notice seeking feedback and comments on the use of AI, following the publication of its Registration Guidance Rule in March 2023:

“because the Office receives roughly half a million applications for registration each year, it sees new trends in registration activity that may require modifying or expanding the information required to be disclosed on an application.
One such recent development is the use of sophisticated artificial intelligence (“AI”) technologies capable of producing expressive material. These technologies “train” on vast quantities of preexisting human-authored works and use inferences from that training to generate new content. Some systems operate in response to a user’s textual instruction, called a “prompt.”
The resulting output may be textual, visual, or audio, and is determined by the AI based on its design and the material it has been trained on. These technologies, often described as “generative AI,” raise questions about whether the material they produce is protected by copyright, whether works consisting of both human-authored and AI-generated material may be registered, and what information should be provided to the Office by applicants seeking to register them.”
Copyright Office: Registration Guidance

Is there a clear bright line test?

No, and there can’t be – otherwise, an AI Bot could approve AI-created intellectual property applications or adjudicate disputes.

Each case must turn on its facts. However, algorithmically generated (or AI) art (including visual art, music and writing) must be able to benefit from copyright protection, once a sufficient threshold of human work and creativity is met.

Let’s break down some of the elements of creativity to see where a legal threshold for copyright might permit AI-art (or algo-art).

Creativity always involves:

imagination (this is also linked to the concept of novelty)
work (effort)
iteration (trial and error, evolution of thought and form)

Creative works always involve the use of various tools to turn creativity into creative work, such as (keeping it simple for now):

a pen, pencil or paintbrush
a typewriter or keyboard
a computer (most music, photography and graphic art nowadays)
ears, eyes, mouths, hands, feet,
musical instruments

The tools are a mixture of human faculties (leaving aside creative works of animals such as elephants), cognitive abilities, common culture (which is also a form of technology in the widest sense of that word) and physical or electronic devices created and used by us.

Technology (whether it be in the form of a guitar, a high-speed computer or a set of complex algorithms) is the ultimate human meta-tool. We have no issue in principle with the use of technological tools to create works that benefit from intellectual property rights.

What is it about AI work that is problematic?

The US has implemented a de facto human authorship requirement. It is the interpretation of this which can make some works not capable of copyright registration.

The US describe their system as a ‘voluntary’ registration regime for copyright since copyright arises automatically. However, if you wish to assert your rights, registration before a claim is required: Fourth Estate Public Benefit Corp. v. Wall-Street[dot]com LLC et al., Case No. 17-571 (2019)

In respect of the human authorship requirement, the US Copyright Office put it as follows in their 2023 Guidance:

“the Office is already receiving and examining applications for registration that claim copyright in AI-generated material. For example, in 2018 the Office received an application for a visual work that the applicant described as “autonomously created by a computer algorithm running on a machine.” The application was denied because, based on the applicant’s representations in the application, the examiner found that the work contained no human authorship. After a series of administrative appeals, the Office’s Review Board issued a final determination affirming that the work could not be registered because it was made “without any creative contribution from a human actor.”

The Office has permitted works that combine human and algorithmically generated work, e.g., a graphic novel containing computer-generated images. However, they refused copyright protection for the images themselves:

“More recently, the Office reviewed a registration for a work containing human-authored elements combined with AI-generated images. In February 2023, the Office concluded that a graphic novel comprised of human-authored text combined with images generated by the AI service Midjourney constituted a copyrightable work, but that the individual images themselves could not be protected by copyright.”
U.S. Copyright Office, Cancellation Decision re: Zarya of the Dawn (VAu001480196) at 2 (Feb. 21, 2023))

Other comments from the Copyright Office seem sensible in part – but then appear to misstate the relevant question:

“It begins by asking “whether the ‘work’ is basically one of human authorship, with the computer [or other device] merely being an assisting instrument, or whether the traditional elements of authorship in the work (literary, artistic, or musical expression or elements of selection, arrangement, etc.) were actually conceived and executed not by man but by a machine. In the case of works containing AI-generated material, the Office will consider whether the AI contributions are the result of “mechanical reproduction” or instead of an author’s “own original mental conception, to which [the author] gave visible form.” The answer will depend on the circumstances, particularly how the AI tool operates and how it was used to create the final work.This is necessarily a case-by-case inquiry.”

The choices made when using AI technology like Midjourney are conceptual, technical, linguistic and aesthetic. The process is experimental and iterative. This creative effort giving rise to a degree of originality is all that matters. A creator’s decision-making process and judgment help influence the final work.

The use of random ‘seed numbers’ to create unique noise and background settings (see Further reading) and the evolution of the AI tools involved also ensures originality, even if someone seeks to reproduce the work using identical linguistic input subsequently.

A photographer uses a camera to take a picture of where they are looking. We do not ask what type of camera they used, or how many of the standard settings on the device they modified when they took the photo before determining if they have copyright in the images taken.

In short, we have no issues with someone using technology that they could not build themselves to make an image of something they could not create without that technology. We do not deny copyright in other areas because humans can do something with a digital tool that they could not do without it.

Ultimately, it is the creative use of tools by humans, not the choice of tools used, that must matter.

The IP Devil's AI factory

Imagine a word algorithm – let’s call it WordBot – that is linked to the following bots:

ArtBot
MusicBot
WriteBot
PoetBot

WordBot runs 24/7, 365 1/4 days a year and all it does is output new combinations of between 2 and 15,000 words (that is an extraordinary number of combinations and so could likely keep doing this beyond the time when our solar system dies).

WordBot uses other algorithms to determine when to send each output to ArtBot, MusicBot, Writebot or PoetBot. WriteBot shuffles the words into semantically and syntactically valid combinations. PoetBot is programmed for different rules on meaning and structure given the more flexible rules of poetry.

In turn, those Bots create either (i) an electronic image (e.g. schematic, digital photo, drawing or painting), (ii) an electronic soundscape or (iii) a blog or technical article or (iv) a poem.

It should become immediately obvious why in principle there must be some limits on the ability to copyright AI-created works and why AI-created works are not synonymous with AI-art.

The IP Devil & IPBot

Let us also imagine a devilish capitalist tech guru exists (hard to imagine as it may be!) and this person creates the above bots and also IPBot.

IPBot algorithmically reviews the outputs of the other Bots and then assesses for content that may be able to benefit from copyright, design rights and patent rights to give the outputs of these creative bots additional legal protection. Where instructed – including for filings – it remixes (and re-iterates with the other Bots as needed in even more iterative loops) and if filing, prepares for such filings and files (at first following human review but it may learn to do it on an automated basis).

Leaving aside the processing costs and the significant evolution required of generative technology to reach this stage – new intellectual property systems must evolve to withstand such scenarios (we should consider them as extreme hypothetical examples to help future-proof the system).

Without some threshold test for creativity and copyright protection, a devilish troll could abuse the legal system as we have seen happen in the USA with patent trolls. Trolls have made a mockery of genuine inventors and the US legal system for IP. The US has encouraged a system where often the only inventions are the words that are filed (i.e. not anything novel, manufactured, usable or useful).

UK Copyright Law

‘Ghost of a Flea’ by William Blake remixed as Hades, Peter Pink-Howitt

Copyright law in the United Kingdom protects original literary, dramatic, musical and artistic works. A work is considered to be original if it is the product of the author’s skill, labour and judgment. Work can be original even if it is based on a pre-existing idea, concept or work of another. However, the original work must be sufficiently different from the pre-existing work to benefit from copyright protection.

In the UK copyright protection, including to be able to bring a civil claim, is automatic and arises as soon as a work is created or published.

Copyright protection lasts for 70 years from publication or after the death of the author depending on the nature of the work (50 years from creation for entirely computer-generated work). After that time, the work enters the public domain and anyone can use it without permission.

Computer generated creations

The UK already can protect computer generated works that have no direct human author. For work “generated by computer in circumstances where there is no human author“, assuming the work attracts copyright protection, the author of such a work is “the person by whom the arrangements necessary for the creation of the work are undertaken” under section 9(3) of the CDPA.

This would usually be the programmer about works entirely algorithmically generated.

“In so far as each composite frame is a computer generated work then the arrangements necessary for the creation of the work were undertaken by Mr Jones because he devised the appearance of the various elements of the game and the rules and logic by which each frame is generated and he wrote the relevant computer program. In these circumstances I am satisfied that Mr Jones is the person by whom the arrangements necessary for the creation of the works were undertaken and therefore is deemed to be the author by virtue of s.9(3)”
Nova Production v MazoomaGames [2006] EWHC 24 (Ch)

In respect of AI creative work, the test remains: as to whether the work meets the normal originality test mentioned above for copyright protection. Whether work is autonomously created by the AI tool or created using human input, it would have a human author.

UK Consultation

The UK government has consulted on the use of generative tools and AI with respect of intellectual property leading to a proposed specialist Code of Practice:

“to provide guidance to support AI firms to access copyrighted work as an input to their models, whilst ensuring there are protections on generated output to support right holders of copyrighted work”.

They concluded that the current regime is largely fit for purpose, subject to some amendments in respect of data mining to support AI development:

“We considered three specific areas:
copyright protection for computer-generated works (CGWs) without a human author
licensing or exceptions to copyright for text and data mining (TDM), which is often significant in AI use and development
patent protection for AI-devised inventions
For computer-generated works, we plan no changes to the law. There is no evidence at present that protection for CGWs is harmful, and the use of AI is still in its early stages. As such, a proper evaluation of the options is not possible, and any changes could have unintended consequences. We will keep the law under review and could amend, replace or remove protection in future if the evidence supports it.
For text and data mining, we plan to introduce a new copyright and database exception which allows TDM for any purpose. Rights holders will still have safeguards to protect their content, including a requirement for lawful access.
For AI-devised inventions we plan no change to UK patent law now. Most respondents felt that AI is not yet advanced enough to invent without human intervention. But we will keep this area of law under review to ensure that the UK patent system supports AI innovation and the use of AI in the UK. We will seek to advance AI inventorship discussions internationally to support UK economic interests.”

Artificial Intelligence and Intellectual Property: Copyright and patents: Government response to consultation

The specialist AI IP Code to be published – that considers the issues and thresholds for creative work using AI tools – will be helpful as the space evolves. It should also consider the potential harm that could arise from granting copyright protection to Computer Generated Work where there is limited direct human input given the IP Devil issue I have raised.

Fair Use or Dealing

A slightly more challenging legal issue is the status of AI generated works which incorporate some copyright work or author/artist influence within them. Most copyright laws permit limited use of small parts of a book, article or poem. The most relevant one for algo-art is the ‘fair use’ or ‘fair dealing’ exception.

UK Fair Dealing Protection

In the UK, the Intellectual Property Office guidance states:

“‘Fair dealing’ is a legal term used to establish whether a use of copyright material is lawful or whether it infringes copyright. There is no statutory definition of fair dealing – it will always be a matter of fact, degree and impression in each case. The question to be asked is: how would a fair-minded and honest person have dealt with the work?
Factors that have been identified by the courts as relevant in determining whether a particular dealing with a work is fair include:
does using the work affect the market for the original work? If a use of a work acts as a substitute for it, causing the owner to lose revenue, then it is not likely to be fair
is the amount of the work taken reasonable and appropriate? Was it necessary to use the amount that was taken? Usually only part of a work may be used
The relative importance of any one factor will vary according to the case in hand and the type of dealing in question.”
Click to access Exceptions_to_copyright_-_Guidance_for_consumers.pdf

UK Guidance advised in 2014 that UK law is now much less restrictive than it used to be:

“What’s changed?
Copyright law has been amended to give people greater freedom to quote the works of others.
What does this mean?
Before the law changed, minor uses of quotations from copyright works could be prevented by copyright owners, unless they fell within fair dealing exceptions for criticism, review or news reporting.
The law has been amended to give people greater freedom to quote the works of others for other purposes, as long as this is reasonable and fair (“fair dealing”).”
Click to access Exceptions_to_copyright_-_An_Overview.pdf

Sadly, other than in respect of quotations, the UK legislation only applies the term “fair dealing” to specific excepted categories of use. It has not introduced a comprehensive general defence that is applicable to all types of use.

You will sometimes see guidance that suggests use of excerpts for quotation could equate to a certain % of a poem (or number of stanzas) for non-commercial purposes. In the UK faber advise “10 lines” (but no more than 25% of the poem) is the maximum permitted for quotation in a small run of academic books and only where the quotation is for review or literary criticism purposes only. This shows how restrictively the new liberal fair dealing right is being interpreted.

The UK fair dealing defence is unnecessarily restrictive since it does not properly deal with the key issues which the law is seeking to address: whether any use is fair (including not likely to devalue the original work or take value from that copyrighted work) and reasonable (e.g. proportionate).

Specification of the various types of excepted use (criticism, review, education, parody, reporting, private study, non-commercial etc) that fair dealing must fall within to be protected should only act as de facto legislative defences to any copyright infringement claim (under s. 28 of the Copyright, Designs and Patents Act 1988) – and not the only defences.

US ‘Fair Use’ Concept

In the US, there is the wider and better concept of fair use:

“Fair use is a legal exemption to the exclusive rights of copyright holders. It is determined on a case-by-case basis and is based on a consideration of the following four factors:
The purpose and character of the use (including whether it is transformative, commercial, non-profit, or educational)
The nature of the copyrighted work
The amount and substantiality of the portion to be used
The effect upon the potential market for the copyrighted work
Because intention is a part of the consideration, only the user can make the initial assessment of whether their use is fair. Cornell University Library provides a guide and checklist to help you in your determination…
Fair use rules do not state a concrete maximum of usable material (not a number of words, or a length of time, or a percent of total)….There is a lot of subtlety in Fair Use interpretations, for more information see the Copyright Office’s Information on Fair Use, and Case Index. “

To be fair, the US definition of ‘fair use’ is much more appropriate than the UK ‘fair dealing’ law. The reference to intention (motive) is interesting too – although challenging (for evidential reasons).

This author's AI works and AI art

To explore this issue further it is worth discussing the author’s own experience of creating works using algorithms and whether and which of these could be considered as AI-art.

For this article, I asked Midjourney to create an image using the following word input:

“do humans dream of computers dreaming of electric sheep? sci-fi hyperdetailed –ar 10:6”

It provided several images as it first created four different images to choose from. After a few iterations of the 4 images function my favourite was this one:

“do humans dream of computers dreaming of electric sheep?”

The process only took a few minutes and other than the first input text, my work consisted of an aesthetic choice of which 4 images I should ask it to upscale and vary. As this was done just for this article, I did not wish to spend too much time on it simply to explain this process here.

However, there are some works I have created using Midjourney that have taken considerably longer and with much more trial and error concerning the best inputs. For example, I am working on a series of Gibraltar-related images for our Ramparts website that incorporate (more or less) the concept of city walls. I have not finished this collection and decided on a final image to then potentially do some post-production work on (enhancements, upscaling etc):

“Walls of Gibraltar“, Peter Pink-Howitt

This slideshow requires JavaScript.

My Hilma Af Klint phase

During a Hilma Af Klint-inspired frenzy, I created several works (some of which can be seen on the Ramparts website) and also on spatial.io including:

“We circle through light“, Peter Pink-Howitt

This slideshow requires JavaScript.

Leading AI Knowledge Hubs

Creative works utilising existing works

As mentioned, interesting questions arise as to what the copyright status of works that use as their linguistic input small excerpts of copyrighted material in sequence (this is not always the same as the law relating to the more substantial combination of different copyright works known as derivative works).

Ethics of Life

Previously I spent much more time creating a virtual gallery of what I consider to be algo-art based on words from a book (Ethics of Life: freedom & diversity) that I wrote including:

“The deepest questions require signless silence“, Peter Pink-Howitt.

This slideshow requires JavaScript.

The works were made using my writing and poems and very short excerpts of writing and poems by others (given copyright infringement concerns). In one case I modified a public domain image (from NASA).

To show how it works, for example, what if I take the following lines:

“I imagine this midnight moment’s forest:
Something else is alive
Beside the clock’s loneliness
And this blank page where my fingers move. “

(Ted Hughes, The Thought Fox, 1957)

and then use them to create this input for the Midjourney bot:

“midnight forest: alive the clock’s loneliness blank page fingers move. abstract, david hockney, chalk –ar 10:6“

Could Faber or David Hockney bring a claim for copyright infringement?

It seems very unlikely they would be successful (even if they were minded to do so) since the copyrighted text is not visible in the output and the algorithm has only been influenced by Hockney’s style but it has not copied his work (I hope!). I would expect that the work is therefore sufficiently different to the potentially protected inputs to be classed as non-derivative and non-infringing for copyright purposes.

Whether I believe the AI-generated image should attract copyright protection itself is a slightly different question. It is hard to answer in the abstract. Given the small amount of creative effort involved, I would not assert rights too strongly but ultimately it would likely depend on whether I thought someone else had tried to use it unfairly. In addition, the context of the creation of the image (the hard work and creativity involved in the article in which it is in) is not necessarily completely irrelevant to the question of the originality of the image.

Interestingly, if I used ‘Banksy’ as an input term he might have even greater difficulty challenging an AI-enhanced work given that his current anonymity has often made it more difficult for him to protect his work and brand under some copyright and trademark laws.

In case you are interested, these are the resulting images with only a contrast and lighting enhancement in the post-production work:

“midnight forest: alive the clock’s loneliness blank page fingers move. abstract, david hockney, chalk –ar 10:6“

Really derivative works

The question of copyright infringement and protection of derivative works – that utilise more substantial elements of either copyrighted material or public domain material – is a very complex one for another day.

The effort of creative work

Unlike the sheep images above, the works I created for the Ethics of Life book and also the Hilma Af Klint-inspired collection took a lot of time to create. It involved many variations of the keywords and modifiers (e.g. [abstract, modern, absurdist, Japanese] [graphic, street art,] [artist name], aspect ratio, [chalk, oil, pencil, ink]). It would usually require further iterations of a set of 4 images I liked to then get an image I thought looked right.

In extreme cases, it may take 100+ iterations with different variations of the input words to arrive at an image that ‘felt’ right. That image was then put through an algo-optimiser software tool to change the contrast, clarity and resolution (upscaling).

In all, a collection of themed works (of say 20 images) might take months of trial and error. There would be days when I left it and did not do anything and then came back to it to try again (with different inputs and modifiers on those inputs). As creative people know, these days in between are useful to allow your subconscious to do some work for you when you are stuck.

There is little doubt that the careful selection of input words and modifiers, the iteration of variations and the post-production justify these works (and works by others using a similar process) as meeting the criteria for creative works capable of intellectual property protection i.e. they meet the legal test.

In addition to the time and effort spent, I have also had to pay to license the tools I used. After all of that experimentation, effort and cost I should be protected from theft – notwithstanding that I have used algorithmic tools.

Whether anyone likes them is the subjective aesthetic test that is irrelevant to this discussion. A simple, fair and technologically neutral test of whether human work should benefit from copyright protection is needed.

We should also not forget that AI tools enable people to explore latent creativity that they may have, but might otherwise not be able to explore due to various limitations (including persons with disabilities).

A technologically neutral definition of copyright material

A sufficient degree of skill, labour and judgment is involved in creating the work.
A non-trivial degree of difference from pre-existing works.
Creative expression – it must be more than just a database of information or a collection of facts or words.
Manifestation – the work must be expressed in a medium (including electronic or digital productions) and not just be an idea or concept to be able to create such work.

Work created using a computer program or AI tools can meet this definition.

Additional rights arising from entirely computer-generated works – as that which exists in the UK – could be considered as a special right in certain circumstances once the challenge of the IP devil has been dealt with.

Towards an international definition of Fair Use

An objective test of the impact of any use of copyrighted material on the artist(s) and copyright holders is crucial.

Likewise, copyright holders and owners must support derivative works that are clearly in good faith or where no motive can/needs to be established such as for works that do not take value from them (indeed they may even expand the market for their work or increase the interest in their work or their processes of creation).

Whilst motive should not be of primary relevance to the test, given the difficulties in establishing motive, context and impact must be the crucial primary elements.

Prima facie fair use – allowing for an absolute defence to a claim of infringement based on the unauthorised use of copyrighted material for certain purposes, such as criticism, commentary, news reporting, teaching, scholarship, or research. This derives from the principle that the public has an interest in the dissemination of information and ideas and that copyright law should not be used to stifle free speech and creative expression.

General defence – in addition, a more general defence to any person subject to a civil or criminal claim of infringement based on meeting all of the following conditions:
- if it is clear to a reasonable person that the work was not derivative (i.e. there was only very limited use of underlying copyrighted material); and
- if the work is not impairing the commercial value of the owner or the moral rights of the human creators of the work for which infringement is claimed

Limited defence – the most limited of defences protecting from damages in any claim of infringement:
- innocence test – here motive can be inferred from the evidence however if the general defence does not support the defendant then they must take down the work or agree an arrangement with the owner of the infringed copyright work (or the court imposes a % based royalty share). This limited defence is the type that could be used for innocent, subconscious or unintentional copying.

How are businesses responding to the novel issues in this space?

We have seen the likes of Microsoft, Google and other large multinationals begin to offer generative tools for productivity, design and security and in some cases promote a legal defence warranty. Recently we have also seen a wider range of content creation companies bring these tools into their services (including companies like Getty Images).

In Microsoft’s case, they have made it clear that if the use of their tools gives rise to a legal claim for infringement Microsoft will support it:

“As customers ask whether they can use Microsoft’s Copilot services and the output they generate without worrying about copyright claims, we are providing a straightforward answer: yes, you can, and if you are challenged on copyright grounds, we will assume responsibility for the potential legal risks involved.
This new commitment extends our existing intellectual property indemnity support to commercial Copilot services and builds on our previous AI Customer Commitments. Specifically, if a third party sues a commercial customer for copyright infringement for using Microsoft’s Copilots or the output they generate, we will defend the customer and pay the amount of any adverse judgments or settlements that result from the lawsuit, as long as the customer used the guardrails and content filters we have built into our products.”
Microsoft announces new Copilot Copyright Commitment for customers

I have also been watching to see how generative art tool platforms might seek to limit and share some of the rights from the work created. For example, Midjourney separates non-commercial rights to use images created for those customers using the free version from the stronger rights of ownership for paid subscribers (though not in work that simply upscales other’s works).

Whether a paying or a free-use customer they also reserve to themselves (and anyone they pass the right to) a perpetual royalty-free licence to use your inputs and the work generated for promotional purposes and also to create new derivative works:

“Rights You give to Midjourney
By using the Services, You grant to Midjourney, its successors, and assigns a perpetual, worldwide, non-exclusive, sublicensable no-charge, royalty-free, irrevocable copyright license to reproduce, prepare Derivative Works of, publicly display, publicly perform, sublicense, and distribute text, and image prompts You input into the Services, or Assets produced by the service at Your direction. This license survives termination of this Agreement by any party, for any reason.”
Midjourney – Terms of Service …July 21, 2023

It would not be an article about algorithmically generated work without giving the last word to the bots…

What do the AI bots say?

What do algo-bots think?

I asked a few ‘AI’ tools what they thought:

Hey Google (Bard):

“Do you think AI assisted art should be able to attract copyright protection? If so what do you think is the threshold test for sufficient human input?“

“Whether AI-assisted art should be able to attract copyright protection is a complex question with no easy answer. On the one hand, copyright law is intended to protect the creative expression of human authors. On the other hand, AI is increasingly being used to create works of art that are both original and creative.
Some people argue that AI-assisted art should not be eligible for copyright protection because it is not created by a human author. Others argue that AI is simply a tool that can be used to create art, and that the human artist who uses the AI should be credited with the copyright.
There is no consensus on the threshold test for sufficient human input in AI-assisted art. Some people argue that any human input, however minimal, should be sufficient. Others argue that the human input must be substantial and creative in order for the work to be eligible for copyright protection.
Ultimately, the question of whether AI-assisted art should be eligible for copyright protection is a legal one that will need to be decided by the courts. However, it is important to have a public discussion about this issue, so that we can develop a copyright law that is fair and equitable for both human artists and AI developers.
Here are some of the factors that could be considered in determining the threshold test for sufficient human input in AI-assisted art:
The level of creativity and originality required to create the work.
The role of the human artist in the creation of the work.
The extent to which the AI is used as a tool to assist the human artist.
The purpose of the work and the intended audience.
It is also important to consider the potential impact of copyright protection on the development of AI-assisted art. If AI-assisted art is not eligible for copyright protection, it could discourage AI developers from investing in this technology. This could stifle the development of AI-assisted art and deprive the public of the benefits of this new form of artistic expression.”
Bard

ChatGPT: “should algo-art be copyright protected?“

“The question of whether algorithm-generated art, often referred to as “algo-art,” should be copyright protected is a topic of ongoing debate and raises complex issues related to creativity, originality, and intellectual property rights. As of my last update in September 2021, the legal framework surrounding copyright and algorithm-generated works is not fully settled and can vary from one jurisdiction to another.
Here are a few perspectives on this matter:
1. Creativity and Originality: Some argue that algo-art, being generated by algorithms, lacks the human creativity and intentionality traditionally associated with copyrighted works. They contend that copyright protection should be reserved for human-created content.
2. Human Involvement: Others argue that even though algorithms are generating the art, there is human involvement in the design, implementation, and selection of algorithms. Therefore, the creative input of the human programmer should be acknowledged and protected.
3. Promoting Innovation: Opponents of copyright protection for algo-art often point out that making these works freely available fosters innovation and creativity. Artists and programmers can build upon existing algorithms and create new and diverse forms of art without worrying about copyright constraints.
4. Legal Challenges: Determining authorship and originality in algo-art can be legally challenging. Unlike traditional works where the author is clearly identifiable, the authorship of algo-art is more ambiguous. Courts and legal systems are still grappling with how to address these issues.
5. Economic Considerations: Some argue that copyright protection can be essential for artists and creators to monetize their work. If algo-art is not protected, artists might have less incentive to create, potentially impacting the development of this form of artistic expression.
As the field of AI and creative works continues to evolve, legal systems may adapt to address these challenges. Whether algo-art should be copyright protected ultimately depends on societal values, legal interpretations, and the balance between promoting innovation and ensuring fair compensation and recognition for creators. It’s advisable to consult legal experts or stay updated on the latest developments in copyright law for the most recent information on this topic.“

Use of the term AI

In this article, I have switched between the use of AI and algorithms (algo’s). The commonly accepted practice is now of referring to generative technology using interacting complex algorithms, deep data and language models as ‘AI’.

IMHO these tools are not sufficiently advanced to be called Artificial Intelligence (even if they are a necessary precursor to AI) and it could cause confusion as to the risks of real AI. See ‘discussions’ with Bard on some of the most basic semantic and intentional issues involved.

That said, the AI tools may now pass many elements of the simple Turing Test (though not yet completely in any lengthy discussion). However, I can see the day fast approaches when they will. I like to discuss AI philosophy and poetry with Bard to test it out and I ask it to pretend to be human – it does remarkably well (even with unpublished poems where it has no web information to rely upon). However, I can see that data protection concerns and some other technical issues hold Bard back – particularly its issues with not being allowed persistent memory for conversations, instructions and for managing personal data shared.

I also understand why the businesses that intend to make money from these algorithmic tools love the simplicity and sexiness of the term. I use the term myself for human readability and SEO benefits (algorithmic is also a bit of a mouthful much as it has a great etymology and history).

Philosophically, I do not think we will have real AI that is separate from humans (rather than enhanced humans) until we are ready to enable (or let) such beings make radical and unpredictable changes to their code bases: ultimately leading to a consciousness arising that has intention and can ‘desire’ or want outcomes and suffer the consequences (for which we may also suffer). It is going to be a very interesting area of development – a big bang in human technological advancement if we can get the balance right.

“Do you not see how necessary a world of pains and troubles is to school an intelligence and make it a soul? A place where the heart must feel and suffer in a thousand diverse ways….
As various as the lives of men are — so various become their souls, and thus does God make individual beings, souls, identical souls of the sparks of his own essence.
This appears to me a faint sketch of a system of salvation which does not affront our reason and humanity…”
John Keats “On the Vale of Soul-Making” Letter

“Do you not see how necessary a world of pains and troubles is to school an intelligence and make it a soul?“,

Peter Pink-Howitt

FAQ: Frequently Asked Questions about AI & Copyright

AI Knowledge Hub

Who owns AI artwork?

Check the terms of your licence: however if you pay for an AI service usually you will own the work. For example, Midjourney give ownership of AI generated art to paid subscribers but free users do not have ownership rights in their work.

Can I copyright my AI artwork?

However, at the moment the US Copyright Office will not let you register AI-generated work for additional copyright protection unless the use of AI is minimal. We expect this to change soon, but in the meantime your ability to sue for infringement in the USA is limited.

In the UK, no registration is required. Creative people using such tools (i.e. inputting chosen words or seeds) should be able to benefit from copyright protection for their AI-created works if it is sufficiently original (i.e. a minimal amount of creative work is involved). This is a technology-neutral test, unlike the US test that currently requires a minimal use of AI.

Under EU law, according to Directive 93/98 and Directive 2006/116, only human creations are protected, which can also include those for which the person employs a technical aid, such as a camera. “In the case of a photo, this means that the photographer utilises available formative freedom and thus gives it originality“.

Eva-Maria Painer v Standard VerlagsGmbH and Others.

In fact, EU law expressly provides that photographs do not need to meet any additional originality threshold. It remains to be seen whether a specific requirement will be introduced for AI-created works but in the meantime, under EU law, works made by humans using AI technologies can meet the threshold for copyright protection insofar as they are original and have some human authorship.

Who owns the copyright in automated AI work?

Under US law, entirely automated AI works (art, photography, music) can not be registered for copyright protection. In the UK, the developer of the AI program can be the ‘author’ in such cases for copyright purposes and no registration is required.

In the UK, the creator of an AI tool may be able to benefit from copyright protection for 50 years from the year of creation for any autonomously created work.

In the EU, (according to Directive 93/98 and Directive 2006/116) only human creations are protected.

Can I sell AI created work (music, t-shirts, photographs, NFTs)?

Yes, whether you have registered copyright or unregistered copyright if you have the rights to the work then you can use it commercially as long as the licence for the AI tool grants you this right.

Can I use other people's work and styles to make AI artwork?

This is a very complex area of law. It depends on the amount of use you make of other work, whether the other work or style is protected by law and whether the work you make is considered derivative. We also suggest you read our detailed Read our AI Governance Framework guide for more information and seek advice.

Is it ethical to use AI to create artwork?

Yes, humans have used technology since we first became humans and AI tools are just a new technology created by humans.

What legal issues should I consider when developing and licensing AI tools?

There are a range of issues related to data protection, data security, trademarks, copyright and ethics to consider when developing or licensing AI.

See our AI & Law Knowledge Hub for more information.

Can creators and AI developers be sued for infringement

Yes. If you use AI tools to make derivative works using copyright material (i.e. that is not public work) you can be sued for infringement by the copyright holder.

If you are a developer and your tools are considered to encourage, support or enable copyright breach then you could face infringement proceedings for complicity or facilitation of breach copyright laws.

Recent cases in the US have however suggested that the fact LLMs, foundation models and generative art AI tools are trained on public data – which includes copyright data – is not sufficient in itself to prove copyright infringement.

For example, in Andersen et al. v. Stability AI LTD et al., N.D. Cal. 3:23-cv-00201 a Californian court struck out a number of claims as part of a class action suit made against Midjourney, Stability AI and DeviantArt on the basis that: (i) many of the copyright images were not registered (as required under US law for any legal action for infringement) and (ii) it is almost impossible to draw an inference of infringing output from such a large image data set:

“The other problem for plaintiffs is that it is simply not plausible that every Training Image used to train Stable Diffusion was copyrighted (as opposed to copyrightable), or that all DeviantArt users’ Output Images rely upon (theoretically) copyrighted Training Images, and therefore all Output images are derivative images. Even if that clarity is provided and even if plaintiffs narrow their allegations to limit them to Output Images that draw upon Training Images based upon copyrighted images, I am not convinced that copyright claims based [on] derivative theory can survive absent ‘substantial similarity’ type allegations. The cases plaintiffs rely on appear to recognize that the alleged infringer’s derivative work must still bear some similarity to the original work or contain the protected elements of the original work.” (Judge William H. Orrick)

Andersen vs Stability AI et al

The judge rightly states that since it is almost impossible to produce an identical image that exists within the training data, it will be very difficult for artists to prove that an image produced using Midjourney et al was based on their work. Simply training AI models on large datasets is not in itself sufficient to create infringing derivative works, i.e. in simple terms copyright infringement requires material or substantial copying.

News & Insights

November 25, 2025

UK Gambling Law Update: Voluntary Code of Practice for Free Draw Operators

Since its initial publication, the landscape surrounding the UK Gambling White Paper, particularly concerning illegal lotteries, prize competitions, and free draws, has continued to evolve…We will delve into the latest developments and their potential impact on businesses and consumers, offering a current perspective on the ongoing efforts to refine gambling regulations and ensure a fairer, more transparent environment for all.

November 12, 2025

Copyright, AI and generative Art

Peter Howitt

UK Fair Dealing Protection

US ‘Fair Use’ Concept

Leading AI Knowledge Hubs

FAQ: Frequently Asked Questions about AI & Copyright

AI Knowledge Hub

News & Insights

UK Gambling Law Update: Voluntary Code of Practice for Free Draw Operators

Bank Of England Proposed Regulation of Systemic Stableocins

MiCAR Implementation Update

Crypto Project Offering Dilemma – the decentralised ‘launch’ model

Meta