Art and AI: Court cases are stretching copyright principles

It is not difficult to understand the tension that exists between the new technologies, such as DreamStudio AI, and the desire of artists to have control over their works and to earn money from them. Inevitably cases and/or laws will draw a line between what’s fair and what’s not.

In the last few months, a number of court cases have been filed regarding the use of artworks by tech companies to “train” artificial intelligence tools. These companies use images from the internet to program their AI tools with different moods, themes, and styles. These lawsuits are primarily aimed at Stability AI’s image generator, Stable Diffusion. Getty Images has sued Stability both in the USA and the UK over the alleged use of millions of images from Getty Images’ library to train Stable Diffusion. Getty claims damages in the USA of $2 trillion.

AI and copyright laws are complex. First, there is the question of whether or not the AI tool’s learning process and the results violate copyright. Stability’s cases are a simple example: Getty claims that Stable Diffusion copied their copyright images in order to train the tool. This copying, if unlicensed, would violate any copyright in the images. Stability cannot defend itself since any programming process involves the reproduction of source material, even if it is stored for a very short time. In the US, the issue of “fair usage” could be relevant in Stability’s case. However, this would not apply to the UK.

The question of whether AI-based tools themselves are infringing is a more nuanced one. In practice, it is more likely that this will involve a determination of how much of the original image has been copied. Stable Diffusion is alleged to have been used by artists to create images that are similar in style, even if the actual work was not reproduced. This complicates infringement analysis.

These cases will be closely watched to gain some judicial certainty on these issues. It could take many years for the courts to catch up with current technology and even future technologies.

There’s always a tension between new technologies and the desire of artists to have control over the use of their works and to earn money from them. So, inevitably cases and/or laws will draw a line between what’s fair and what’s not.

There are plans to introduce legislation on the subject of infringement. The UK Government has proposed to expand the exception to the infringement rules for data mining that is currently allowed for non-commercial purposes to allow it for any purpose. This would permit training AI tools without committing infringement.

Another side of the coin concerns the protection of AI-generated works. Can AI-generated work be considered “original” to be protected by copyright? This question can be very hard to answer in the context of AI tools that are able to learn styles and images. It may also require a case-by-case analysis.

Who owns the copyright if it exists in the artwork created? According to the UK copyright law, the owner of the copyright is the original creator or author. The author can then reproduce the work or grant others the right to do so through a license or assignment of the copyright. What if, however, the digital asset was not created by a person but by Artificial Intelligence instead? Who owns a copyright to something that is created by a computer, where the “artistic endeavour” or the “labour, skill, and judgment” of an artist are absent? CryptoPunks is a good example. These are computer-generated images that are all different but have a similar 8-bit format. Who owns these computer-generated images?

The UK’s Copyright Designs and Patents Act of 1988, which was passed in 1988, is unique because it tries to address this issue by referring to “computer-generated works.” The owner is defined as the person making the “arrangements” necessary to create the work. However, this may not be easy to determine, and it can be difficult to pinpoint where and who made the arrangements.

In order to gain some judicial certainty on these issues, the Stability cases are being closely monitored. However, it could take many years for the courts to catch up with current technology. It’s worth noting that AI is affecting many fields and not just art. For example, AI writing tools are also “a threat” to content marketing jobs.