Memes and copyright: it works without Conni too

• Firstly, mass copying, familiar from piracy and file-sharing networks (and, more recently, in a different form through AI training).
• Secondly, automated or user-generated aggregation, familiar from search engines, social media and streaming platforms.
• Thirdly, unauthorised transformation, familiar from the creative possibilities offered by platforms (and, more recently, by generative AI systems).

We will take a closer look here at the third issue: editing and transformation on the social web. A collective management organisation for image rights must provide solutions for this. However, in practice, we at ProLitteris deal almost exclusively with art licences, i.e. professional reproductions of works of fine art and artistic photography. We almost never deal with consumers or prosumers in the context of collective management. Yet we regularly receive such enquiries for legal advice.

At first glance, memes, edits and remixes are just a bit of fun. Sometimes, however, markets and market influences emerge for the management of professional photographs and illustrations. And there are distortions and disparagements that authors, publishing houses and producers are not happy to put up with. Last but not least, the rampant sharing of excerpts from works on social media also means that their dissemination and reception spiral out of control. It will no longer be possible to listen to the popular and often arbitrary pieces of music used in posts on social media and streaming services without prejudice.

Despite these excesses, adaptations also come from fans and friends of the creators and producers of the original content, and they can have a promotional effect – albeit an uncontrolled one. Even if this closeness and appreciation are lacking: who would want to take legal action against consumers and online communities?

Copyright law generally requires consent for adaptations, whilst granting privileges for creativity and criticism. Thanks to these legal freedoms of use, the problem is almost solved:

• In the EU, permitted adaptations are known as ‘pastiches’, which are understood to be a creative engagement that draws on the original but at the same time differs from it and does not adversely affect the original’s market.
• The US ‘fair use’ principle provides a legal exemption for use, subject, amongst other things, to the purpose and nature of the use. Transformative works – that is, original new expressions – strongly support a claim of fair use.
• The Swiss Copyright Act contains a provision on freedom of parody, which also covers comparable adaptations of the work. This refers to satire, parody or criticism of the work or its authorship – or even of the work itself?

This is the crux of the matter, both as a legal issue and as a political issue within copyright law: am I really allowed to use any work to illustrate or reinforce any statement? Is this also the case under Swiss law, where an older legal provision applies than in the EU? Does this freedom mean that I can help myself freely and frivolously to the entire cultural heritage, without regard for the material used and its own meaning and message – as long as I do not impair the market or the management of the original? So, do I take Munch’s *The Scream* to symbolise outrage, Picasso’s *Dove* to symbolise a love of peace, the opening bars of Beethoven’s Fifth to symbolise fate, and film clips of Darth Vader and Voldemort to symbolise evil?

This broad interpretation of freedom means that parody, engagement and adaptation no longer have anything to do with the work being used. The artist, author, filmmaker and so on lose control over how their work is used and its context. The impact and reception of their works are diluted. The spreaders of memes and everyday consumers may not care, but what about the creators?

The problem is growing, as adaptations have become much easier with the new image, audio and video generators. The era of collage snippets is over. Production takes seconds (which is also part of the solution; see below).

It is worth noting that such a broad application of freedoms of use goes beyond the requirement of necessity that characterises other legal privileges:

• Quotation: I use the protected work to the extent necessary for my own expression.
• Reporting: I am reporting on an event in which the image, music, etc. featured.
• Catalogue and directory freedom: Exhibition and auction catalogues or museum directories provide information about existing works.

The example of the children’s book character Conni illustrates what the problem can be. During the summer months, AI-generated memes circulated on social media that used the book character Conni in a humorous, mocking or even hurtful manner, without the consent of the publishing house or the authors and illustrators. For instance, in the face of the first heatwave in early July, the Düsseldorf Fire Brigade warned of the danger on Facebook: ‘Conni doesn’t go swimming in the Rhine’. That is advertising. The meme ‘Conni joins the armed resistance’ is a political message. ‘Conni opens a cotton plantation’ is a racist remark.

When Carlsen Publishing House understandably took action, it sparked a storm of criticism: how could anyone possibly take action against Conni memes! And rightly so: under recent European and German law, rights holders must tolerate memes and similar adaptations as long as they are not commercial or distort the original work.

But is this rule – to which the aforementioned publishing house also had to submit – the final word? Would it not be more appropriate to require that a parody or adaptation relate to the work in question? After all, a free debate should take place in relation to this work, its authorship or its immediate context, not in relation to anything else. For it is only then that constitutional fundamental rights are affected: I must be allowed to display a work (necessity) in order to engage with it publicly. The work should be the purpose, not the means, of the expression.

Furthermore, AI should be the solution, not the problem, as the generation of original images, audio and videos using AI systems has been massively simplified. Whereas one used to have to cut things out or draw them oneself, the software now produces anything quickly, and one can easily instruct it not to copy existing material or characters one-to-one.

The image above – the comic adaptation of the ‘Disloyal Man’ photograph – demonstrates that the specific, recognisable reproduction of elements from copyright-protected material is not necessary. What has been adopted is the triangular relationship of gazes (idea) and the Ghibli aesthetic (style).

To spell this out in more detail for fans of image rights:

• The illustration in question does not show the photograph itself, but rather a variation that does not incorporate the individual characteristics of the photograph, for the purpose of examining copyright rules.
• Due to its composition and design, the photograph is an individual intellectual work. A licence is available as a stock image.
• The common practice of superimposing text onto the photograph and subsequently making it available online (sharing) constitutes multiple uses subject to licensing – though in many cases this is tolerated as a ‘parody of something’.
• Creating and sharing an AI image in the style of the Japanese film studio Ghibli is permitted as a stylistic imitation, provided that no specific individual design elements are copied. The creators and distributors of the AI image therefore do not need to obtain a licence for Ghibli’s copyright. This is because what is being done here is not a reproduction or adaptation subject to licensing, but a licence-free stylistic imitation. Legally speaking, Ghibli could only take action against memes that use a specific character or design.
• The AI image itself is not subject to any rights; nor is it a photograph, as it is computer-generated.
• One possible argument is that the original photograph ‘Disloyal Man’ remains recognisable as an intellectual work in its individual characteristics, with the consequence that a licence would be required in favour of the photographer or their stock image platform, even for the Ghibli recreation. However, in this case, what cannot realistically be prevented is likely to be tolerated.
• The image of the young man turning to look at a woman walking past, whilst his girlfriend stares at him indignantly, is used millions of times on the internet. The concept of the three people is, of course, in the public domain – but one must either refrain from using the image or alter it significantly.
• Rights of personality: According to the photographer, the models have waived their rights of personality in relation to this image (and were presumably paid for doing so).

A practical guide to memes might therefore read as follows:

For adaptations made in the exercise of fundamental rights of communication (freedom of expression, artistic freedom, academic freedom, media freedom), copying ideas and styles is legally permitted and unobjectionable. The reproduction of specific works or parts thereof is permitted by law if it is necessary and if the expression relates directly to the material used or its authorship, namely as a parody or criticism. Otherwise, the permission of the rights holders is required.

This guidance seems to us to apply to memes, even though international law and, depending on one’s interpretation, Swiss law and legal practice are more tolerant when it comes to the internet. More extensive use of protected works and creations should be based on licences – these could also be collective licences with flat-rate royalties, as is the case in other areas.

Back to Conni. As a copyright-protected work, she should not, strictly speaking, have to be used for online content intended to make a point or a joke that has nothing to do with Conni.

The solution is called Bonni. The design is similar, but not identical: it copies the idea and style, but is not a direct copy and does not feature any recognisable individual characteristics of the original. If those adapting the work do want the original – Conni – it is probably because the recognisability and credibility are an advantage – in other words, the intellectual property itself.

Philip Kübler created this image as an example using ChatGPT. It is not subject to copyright, as a machine-generated work is not an intellectual creation; the prompt was trivial, and the image contains no individual intellectual creations derived from pre-existing works. It took a matter of seconds.