• Siobhan Ali

Art Pot x Law Review: Disputes Surrounding the Use of Artificial Intelligence in Art

Over the last several years, Artificial Intelligence (AI) has permeated various aspects of our existence, from healthcare and retail to law and security. For example, in 2018, the first AI-generated painting to be auctioned by Christie’s sold for almost half a million dollars. The “Portrait of Edmond de Belamy” was crafted by feeding images of 15,000 portraits dating between the 14th and 20th centuries into an algorithm. The generated image was then printed onto canvas in ink, signed with the mathematical formula used to generate it. Similarly, the New Rembrandt” features fragments from nearly 170,000 Rembrandt paintings to create a 3D print incorporating the painter’s use of “geometry, composition and painting materials”. Bringing the icon to life once again, the algorithm aims to engage with his work and conduct further research into his artistry.

This expansion of AI into creative fields such as literature, music, and art has brought several legal problems and concerns to the forefront. In particular, it has blurred the lines regarding Intellectual Property (IP). Related to the “creations of the mind”, IP offers legal protections like copyright, trademarks, and patents which protect a range of creative works such as images, names, inventions, and texts, to name a few. It has also brought into question key terms, from “author” and “artist” to “inventor”, “creator” and the like. For example, Hugo Caselles-Dupré, co-founder of “Obvious”, a Paris-based art collective which generated the Portrait of Edmond de Belamy, explains how their project parallels the “mechanics within the human brain” with those of an algorithm. In this way, it plays with the concept of the creativity of a machine.

Historically, copyright has protected the “human creative spirit”. Most jurisdictions’ copyright laws, therefore, do not extend to non-human authors. Beyond moral arguments that argue such provisions fail to sufficiently protect artistic works, the failure of AI-generated art to qualify for copyright could also have disastrous commercial consequences. The core foundation of IP systems is to stimulate technological growth and creative development by creating favourable economic conditions. AI-generated creative works require significant capital. If their products are deemed free for reuse by third-parties, programmers would not reap any financial benefit. Companies would therefore not be incentivised to produce such work, causing a decline in investments in automated systems.

One viable solution to this issue is attributing authorship of AI-generated works to the individuals who created the programme. According to Section 9(3) of the United Kingdom’s Copyright, Designs and Patents Act 1988, for any:

“literary, dramatic, musical, or artistic work which is computer-generated, the author shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken”.

This identifies the “human behind the artificial process”, awarding them rights over the product. In this way, it acknowledges the labour, artistry, and intellect of the individual whose creative ideas formed the basis of the computer’s work. This also incentivises artists and companies to invest in such media as they are guaranteed a return on their efforts.

Technology has long been involved in creative processes, with computer-produced art dating back to the 1970s. During this period, it was more of an “instrument or a tool very much like a brush or canvas”. That is, it merely assisted a person in bringing his or her artistic vision to fruition. The 1988 Act therefore best serves this function of AI in the artistic process.

However, modern art generated with machine learning technology is no longer dependent on a programmer’s creativity. Instead, the machine acts autonomously, receiving inputted data and making its own independent decisions on what form the product should take. This is described as a “neural network” and is much like the thought processes carried out by humans. The user in question may do little more than simply pressing a button while the machine carries out the bulk of the labour. Moreover, it is predicted that as technology advances further and machines are fed greater datasets, it will be “better at mimicking humans” and may require little to no intervention on their behalf. Why then is ownership of this work granted to an individual who has made minimal contributions to the finished product?

Moreover, it is unclear who would be responsible for the necessary “arrangements” to produce such work. Would this be the programmer who designed the algorithm or the user inputting the data? Andres Guadamuz, Senior Lecturer in Intellectual Property Law at the University of Sussex, compares this to the “maker of a pen or the writer” and highlights how developers behind programmes such as Microsoft Word do not own every piece of work produced by the software.

However, awarding AI intellectual property rights would also make it liable to lawsuits regarding infringements and hold it accountable to legal contracts. For all intents and purposes, AI would, therefore, “be treated as a human for legal process”. This opens up a broad array of questions regarding whether AI machines can be classified as legal entities.

With the core objective of copyright resting on its ability to protect creative works, its failure to do so adequately in regard to AI-generated content defeats its purpose entirely. Nevertheless, copyright laws have faced significant reform over the last several decades and expanded to encompass a range of art forms from the originally protected literary works to plays and productions. Therefore, hope remains that there will be greater engagement with Artificial Intelligence in the field of art law.


This article is part of a collaborative series with publication Art Pot which aims to show how law and art have, and continue to be, interconnected. Read Art Pot's article on the use of Artificial Intelligence in art here.