Zarya of the Dawn – copyright and an AI

Those of you who heard Shira Perlmutter, the US Register of Copyright, on her Australian tour last year will recall the US Copyright Office had withdrawn and was reconsidering the copyright registration for Zarya of the Dawn.[1] On 21 February 2023, the US Copyright Office announced the outcome of that review. While the Copyright Office allowed registration of some aspects, it rejected the claim to copyright in the images created by Midjourney, an AI.

Image of a young feminine looking person with golden skin, dark brown eyes and dark brown hair in corn rows - against a teal background
Zarya of the Dawn – Cover Page

The work(s)

Zarya of the Dawn[2] is a comic consisting of images and text depicting Zarya’s adventure to different worlds to collect mental health tools to handle their emotions, thoughts as a non-binary person.

A page with 4 comic images; the first of which is a postcard of some otherworldly place. The young, feminine looking person reads the card addressed to 'My Dearest Zarya'. They wonder if they are Zarya and why they cannot remember their name
Page 2 of the Zarya of the Dawn comic book

The Copyright Office accepted that the applicant, Ms Kristina Kashtanova, was the author of both the text and the selection and arrangement of the text and images. However, the Copyright Office refused registration for the images themselves on the grounds that they were generated by Midjourney and did not have a human author.

How Midjourney generated the images

As described by the Copyright Office, Midjourney generates an image in response to instructions (called “prompts”) input by the user. The Copyright Office illustrated this process by the prompt:

/imagine cute baby dinosaur shakespeare writing play purple

which generated the images below:

4 images generated by Midjourney of purple coloured, baby dinosaurs each holding a pen and working over a manuscript. In 2 images, the dinosaur looks to the right bottom corner; in the other two, to the right bottom corner
Baby dinosaur writing a play

The user could click on the blue “recycle” image to generate four new images. The user could also refine the images regenerated by providing URLs of images to be used as models or by providing more detailed instructions.

This was not authorship for copyright purposes

For copyright to subsist in original works such as text (literary works) or images (artistic works), US law, like Australian law (see further below), requires the work to be original. That requirement in turn requires the work to be made by a human who is an author. And, according to the Copyright Office, an author is the person “who has actually formed the picture,” the one who acts as “the inventive or master mind.”

At least in theory, if someone gave a draftsperson sufficiently detailed instructions about what a drawing should depict, they rather than the draftsperson may be the author.[3]

The Copyright Office, however, found that the instructions Ms Kashtanova gave to Midjourney did not make her the author of the resulting images. This was because it was not possible to predict the outcome resulting from her prompts:

A person who provides text prompts to Midjourney does not “actually form” the generated images and is not the “master mind” behind them. Instead, as explained above,[4] Midjourney begins the image generation process with a field of visual “noise,” which is refined based on tokens created from user prompts that relate to Midjourney’s training database. The information in the prompt may “influence” generated image, but prompt text does not dictate a specific result. See Prompts, MIDJOURNEY, (explaining that short text prompts cause “each word [to have] a more powerful influence” and that images including in a prompt may “influence the style and content of the finished result”). Because of the significant distance between what a user may direct Midjourney to create and the visual material Midjourney actually produces, Midjourney users lack sufficient control over generated images to be treated as the “master mind” behind them.

The Copyright Office recognised that additional prompts could be applied to initial images to influence subsequent images, however, the process was not controlled by the user as it was “not possible to predict what Midjourney will create ahead of time.”

The Copyright Office contrasted the way Midjourney works with the way an artist might use Photoshop or other tools:

The fact that Midjourney’s specific output cannot be predicted by users makes Midjourney different for copyright purposes than other tools used by artists. See Kashtanova Letter at 11 (arguing that the process of using Midjourney is similar to using other “computer- based tools” such as Adobe Photoshop). Like the photographer in Burrow-Giles, when artists use editing or other assistive tools, they select what visual material to modify, choose which tools to use and what changes to make, and take specific steps to control the final image such that it amounts to the artist’s “own original mental conception, to which [they] gave visible form.”15 Burrow-Giles, 111 U.S. at 60 (explaining that the photographer’s creative choices made the photograph “the product of [his] intellectual invention”). Users of Midjourney do not have comparable control over the initial image generated, or any final image. (emphasis supplied) (footnotes omitted)

Ms Kashtanova also contended that her modifications in Photoshop to some images constituted authorial contribution to support her claim to copyright. From the description in the Copyright Office’s decision, some of the work seems more like touching up or editing rather than authorship. As the material before the Copyright Office did not include the “before” and “after” images, however, the Copyright Office was not include to accept those claims either.

An Australian perspective

Australian courts have also ruled that an author must be a human. Applying the IceTV case, the Full Federal Court has ruled that the processing of telephone subscriber name, address and phone number details into a directory by a computerised database did not qualify as an original copyright work as there was no human author. In the first Telstra v PDC case, Perram J explained at [118] – [119]:

The Act does not presently deal explicitly with the impact of software on authorship (although this is not so in the United Kingdom: s 9(3) Copyright, Designs and Patents Act 1988 (UK)). But a computer program is a tool and it is natural to think that the author of a work generated by a computer program will ordinarily be the person in control of that program. However, care must taken to ensure that the efforts of that person can be seen as being directed to the reduction of a work into a material form. Software comes in a variety of forms and the tasks performed by it range from the trivial to the substantial. So long as the person controlling the program can be seen as directing or fashioning the material form of the work there is no particular danger in viewing that person as the work’s author. But there will be cases where the person operating a program is not controlling the nature of the material form produced by it and in those cases that person will not contribute sufficient independent intellectual effort or sufficient effort of a literary nature to the creation of that form to constitute that person as its author: a plane with its autopilot engaged is flying itself. In such cases, the performance by a computer of functions ordinarily performed by human authors will mean that copyright does not subsist in the work thus created. Those observations are important to this case because they deny the possibility that Mr Vormwald or Mr Cooper were the authors of the directories. They did not guide the creation of the material form of the directories using the programs and their efforts were not, therefore, sufficient for the purposes of originality.

The consequence of those conclusions is that the directories were not copied from elsewhere but neither were they created by a human author or authors. Although humans were certainly involved in the Collection Phase that process antedated the reduction of the collected information into material form and was not relevant to the question of authorship (other than to show that the works were not copied). Whilst humans were ultimately in control of the software which did reduce the information to a material form, their control was over a process of automation and they did not shape or direct the material form themselves (that process being performed by the software). The directories did not, therefore, have an author and copyright cannot subsist in them. (emphasis supplied)

See also Yates J at 169.

This appears to be consistent with the approach taken by the US Copyright Office although both Perram J and Yates J recognised that whether some particular claimed work falls on the “copyright” or “not copyright” side of the line is a question of judgment and degree.

Zarya of the Dawn (Registration # VAu001480196)

  1. No, as I am sure you know, you do not have to register your claim to own copyright in Australia. Registration of copyright is just one of the way Americans are different to most of the rest of us. In Australia copyright comes into existence automatically by the act of creating the material (and not slavishly copying it from some pre-existing material). There is no need to register it. Who the owner of the copyright is will depend on a number of factors such as the type of material – a literary or artistic work or an audio-visual work such as a film or a sound recording or broadcast; whether or not the work was made in the course of employment and whether there has been a written assignment or other contractual arrangement. (That is the sort of thing that requires advice based on the specific individual circumstances.)  ?
  2. This is a link to the donationware download but the Copyright Office’s decision includes images of the cover and page 2.  ?
  3. While the creation of an artistic work raises rather more challenges, an obvious illustration of this theory is the case of someone who dictates a letter or a book to an amanuensis.  ?
  4. Earlier the Copyright Office had explained: ‘… Midjourney “does not understand grammar, sentence structure, or words like humans,” it instead converts words and phrases “into smaller pieces, called tokens, that can be compared to its training data and then used to generate an image.” … Generation involves Midjourney starting with “a field of visual noise, like television static, [used] as a starting point to generate the initial image grids” and then using an algorithm to refine that static into human-recognizable images.’  ?
%d bloggers like this: