top of page

Artificial Intelligence, Art and Copyright Law

Aritifical Intelligence has already shown itself to be capable of many basic and advanced tasks, and is raising questions around what its development may mean for the job market. While many welcome the new technology as an effective tool for increased productivity, many are concerned that the increased use of AI will lead to (mass) unemployment, or may have detrimental effects for occupations for which quality requires a human touch. One market in which AI has become a particular concern is that of the arts and artistic creativity.

While the field of art and the creativity it involves are often considered very personal, subjective, emotional and therefore requires a human touch, the use of generative AI has already been making vast steps to close this supposed gap. Artwork generators like DALL-E, Stable Diffusion and Midjourney have already shown themselves to be capable of creating artwork that is convincingly human and may be used commercially. This will continue to advance as these technologies improve and develop. Outside of the visual arts, AI continues to grow more advanced in the literary field, with programs like ChatGPT able to engage in both creative and academic writing endeavours.

In the face of this sudden advance in technological progress, many artists are concerned about the field of human creativity. These are concerns that have to be addressed by law, particularly in regards to copyright. These concerns largely fall around how we should treat art created by AI and how we should treat the art used to train and inspire AI. There are also further concerns from artists surrounding the ability of AI to create art in a sufficiently meaningful, emotional or expressive way.


The first question surrounding how we treat art created by AI has been answered by many, including by the U.S. Copyright Office, who has used existing copyright precedent to respond to this new technology. Though AI can be used to create art, the U.S. Copyright Office states that AI cannot hold copyright without "significant and creatively transformative human contribution" to the art. This decision is primarily influenced by precedent that the work needs to be an original work of authorship, for which the author must be human. This precedent was set when PETA sued for the rights of a monkey named Naruto to gain royalties for selfies he took with a researcher’s unattended camera. The monkey was found to lack the authorship required for copyright. This precedent further works against the creative work of any machine.


However, even though AI cannot produce any copyrightable artwork itself, human engineered works that incorporate AI may still be copyrighted. The precedent for this was made primarily through a decision on the copyright of a comic book entitled Zarya of the Dawn, whose writing was entirely done by Kris Kashtonova, but whose art was made through the art generator Midjourney (with some minimal editing). While existing precedent denied copyright protection to the images themselves, the book was still given copyright for Kris Kashtonova’s written contribution. The Copyright Office has since released Copyright Registration Guidance for Works Containing Material Generated by AI, cementing the precedent as policy. Though this limits copyright’s application to human works created with AI, some believe that copyright for works solely created by AI may eventually be fully legalised, given that the ability to create copyrightable material with AI could be an advantage. An example of this may be found in China, who has given copyright to news articles created by AI.

This movement to restrict copyrights for works created by AI creates benefits for people on both sides of the debate, with AI users able to utilise its creations, while still requiring an artist’s creative contribution if they wish to hold copyright. Yet a further issue has been found in the method of AI art generation, which works primarily by imitating and learning from large stores of existing art. The way AI utilises this art has become problematic for many artists, whose style can be replicated by these works without any royalties or payments made to them. Though many have accepted AI cannot hold copyright, this does not necessarily restrict its commercial use, nor the use of other artist’s works in making such commercially used products.

Art made in certain styles can be requested from AI. These often explicitly use the name of the artist(s) the style imitates, which allows users to gain commercially usable material in the style of an artist without actually paying the artist themself. The difficulties in this lies more in the intricacies of fair use, for which further precedent may be required before a conclusive claim is made. Yet the argument on behalf of artists seems far more difficult to make in this more nuanced case, with many believing the answer must lie outside copyright law. While it is clear that artists may deserve some kind of compensation, style is not something that can be copyrighted. Just as an art piece made in the style of Van Gogh will still hold copyright, provided it doesn’t imitate any singular work too closely, original art made by AI –even those that are stylistically inspired by existing pieces– does not violate any copyright. However, an alternative solution by the right to publicity may address artists’ concerns, essentially arguing on the grounds that your image may not be exploited promotionally without your permission. Artists may find that taking art in their style without permission makes them appear to associate with and be promoting something they do not, giving them a legal route that many found applicable looking at the song Heart on my Sleeve, whose AI vocals were made to imitate Drake.


Fortunately, in the face of the legal and ethical ambiguity presented by this use of artist’s work in generative AI, many systems have begun to move in a direction that minimises this concern. Adobe has begun integrating a generative AI system that only utilises images for which Adobe holds the copyright, and many other companies are now retraining their AI in a similar manner. While this issue is by no means resolved, it is clear that legal systems across the world are all having to adjust to this new development in AI, and we can only hope that the field of art is well represented as it faces this new technological age.

Comments


bottom of page