The advancement in artificial intelligence (AI) has resulted in generative AI (GAI), capable of creating media, images, and writing. GAI’s ability to create and produce artistic outputs, puts forth the question, ‘Who owns the artwork generated by AI?’ In this paper, I will argue that the user of GAI models is the owner of the GAI’s output. It may be argued that granting these users ownership over this form of artwork is unjust, and that the true ownership over AI generated art, lies with the GAI’s creators, the artists whom the AI took its data from, the GAI model itself, or perhaps to no one at all. I will refute these claims and explain how assigning ownership to GAI users is the most ethical option. For these individuals should uphold ownership and moral responsibility over the artwork which reflects their ideas.
To begin, I will define what is meant by generative AI, artwork, and ownership within this paper. GAI is a form of machine learning, “that can produce content such as audio, text, code, video, images, and other data.”[1] Furthermore, artwork is a form of expression, utilised by people to illustrate their experiences, thoughts, and ideas into words, pictures, and sounds. To define further, “[a]rt is very wide. Art is not only what you see with your eye, but it is within your mind also.”[2]Thus, art extends beyond visualisations, and rather, acts to reflect the human mind in various mediums. Moreover, in this paper, ownership is defined to be a form of authorship, where one has the legal and moral rights over an asset – to which they are held accountable for.
Individuals who create art through GAI, become the owners of the GAI’s output(s). Indeed, if a user creates a piece of artwork through GAI, they are thus, the artist of the artwork produced. This is because the user is utilising GAI as a tool to create art, which is generated based on the prompts and lines of information that they give to the GAI. Because these outputs reflect the visions, ideas, and creativity of the GAI’s users, it would thus, be unfair to claim that these users lack ownership over these outputs. Certainly, it would be unfair to not credit someone if the output in question (whether that be artwork, a business, or a project) were to consist of their ideas, creativity, thoughts, etc. By establishing that the users who generate artwork through GAI are, therefore, the artists of these artworks, this ensures that these individuals uphold ownership over products (namely AI art) which reflect their originality.
Conversely, many disagree with this view, and argue that ownership should reside elsewhere. It may be argued that the designers of the GAI models, are the owners of the GAIs’ outputs, for they created the GAI. I disagree with this claim, for the creators of the GAI model(s) did not contribute to the creative process of the GAI art, but merely created the tool which was used to generate this art. Indeed, it would not be right to claim that the owner of Faber Castel for instance, owns all the drawings made with their tools and materials (such as pencils, markers, etc). The true ownership of these drawings would reside with the artist who used Faber Castel’s materials, for the work produced reflects the artist’s imagination and vision. This point is further supported by Cetinic and She, who state that “the idea that developers of a computational model are entitled to authorship rights can perhaps be dismissed in the same way in which it would be absurd to give credits for artistic photographs to the inventor of the camera”.[3] Perhaps these examples seem too farfetched, but the argument remains the same. The designers of GAI models are not the owners of GAI art, for they merely created the tool which is used by GAI users to create art.
Furthermore, it may be argued that granting complete ownership to the GAI users is unfair, for GAI models utilise art from other artists. Certainly, many believe that some ownership should be given to the artists whose work contributed to the AI generated art. As argued by McCormack et al., “authorship of AI generated art [should] be shared between the artist, the developer of the underlying algorithms, and possibly any number of contributors to the training data whose style is being abstracted.”[4] While I believe that the contributors of the artwork generated should be acknowledged, I do not believe that they should hold any ownership over the GAI’s outputs. This is because these artists do not own the elements in which the GAI takes inspiration from. For instance, it is permissible for A to take inspiration from B, for B does not have ownership over the art styles and elements which inspired A.
Furthermore, ownership should not be granted to various contributors, for these contributors are not involved in the process of creating the AI generated art. To explain, if artist A were to take inspiration from artist B, this does not mean that artist B owns the work in which artist A creates from this inspiration. This would be unfair, for artist B did not contribute to the creation process of the art, but merely inspired its outcome. Much like artist A, users of GAI are not entitled to give, nor share, ownership with artists such as artists B (various contributors), as they merely inspired the final output. Therefore, GAI art should be owned by the GAI user, for while the artwork produced takes inspiration from various artworks (much like every other piece of art), it is nevertheless representative of the user’s ideas and creativity.
However, the following questions arise when such ownership is granted purely to the GAI users: What if the user puts in the prompt ‘generate me artist’s B’s painting X, in style Y? Are they still the owner of the artwork produced?’ While I have argued that the users of GAI are the owners of the GAI’s outputs, they should only be considered owners if the artwork produced is new, original, and does not plagiarise the work of others. There is a difference between taking an aspect from one’s work (e.g., replicating a painting style) and plagiarising another’s work entirely. By generating artist B’s work with slight changes, the user in question has plagiarised upon artist B’s work. The work, although in a different style, nevertheless expresses the ideas, interests, and creativity of artist B. Due to this, ownership over these works do not belong to the GAI user, but rather to artist B. Therefore, while the GAI user ‘created’ a replica of another’s artwork, they do not own this replica and hence, cannot use this artwork for commercial purposes. However, copyright regulations are already in place to prevent the commercial use of plagiarised works, preventing the sale of art if it is too similar to that of the original artwork.[5] Given this, such concerns can be resolved, for ownership will not be given to artist who have copied art which already exists.
Moreover, “[s]ome people say that the AI system itself could be considered the creator and owner of the copyright, just like an artist would own the copyright to their work.”[6] I disagree with this view, and argue that AI is incapable of owning artwork, for the art produced consists of human involvement. Without the users’ inputs, the GAI would be incapable of producing art. Given the model’s dependency on human ideas to generate artistic outcomes, the argument that ‘AI creates its own art’ is misguided.[7] Ownership of such works cannot belong to these models, for the artwork produced expresses the ideas of the GAI users, which the GAI models require to generate outputs.
Thaler, a US computer scientist, disagrees with this claim, and urges that GAI models should own the work in which they produce as they are sentient. Thaler argues that “his AI programs are sentient and should [thus,] be legally recognized as the creators of [the] artwork … they generate”[8]. Thaler argues that his GAI, DABUS, is sentient as it is “a system, with … subsystems governed by their own particular algorithm … [much] like the brain”.[9] While I can agree that a fully sentient and conscious machine, should be considered the owner of the works it produces, such machines currently do not exist. Certainly, if a GAI model were to create art, without human input which expresses its own interests, thoughts, and experiences, then it should rightfully be considered the artwork’s owner.
However, such machinery does not exist.[10] Henceforth, while Thaler claims that “we already have sentient AI, we don’t.”[11] “[T]hese systems are not autonomous … as humans provide significant contributions to the very system that leads to the inventive output.”[12] Given this, ownership over AI art should remain with the GAI users, as the art produced reflects their thoughts, ideas, and visions.
However, some individuals disagree with all these stances, and argue that AI art is owned by no one. Zurth provides an example of this view, and argues that neither the designers of the GAI model, nor its users, should have ownership over AI generated art. Zurth explains that neither the creators nor users of GAI models, have enough input in the AI art to be considered its owner(s).[13] He argues that AI-generated works lack the necessary human creativity and originality to warrant copyright protection, and that granting these rights to AI systems can result in a legal and economic imbalance. I disagree with this position and argue that ownership should be assigned to the users of GAI, as these users are morally responsible for the AI art they produce. For instance, if a user were to give the AI an unethical prompt, for example, one in which instils racist stereotypes, it is only fair for the user to be held accountable for the GAI’s racist outputs. This is because the GAI’s output was based on the user’s racist prompts and ideas. If the artwork were to be owned by no one, who would take accountability for these controversial and immoral outputs? It is only right for this responsibility to remain with the GAI users, for the art produced nonetheless expresses their ideas. Regardless of whether the users’ input was insufficient in Zurth’s eyes, GAI users should be given ownership over these artworks, for they should be held accountable for the artwork(s) which reflect their ideas.
Overall, artwork generated through GAI should be owned by the GAI users who created the artwork through their prompts and creativity. Assigning ownership to these users is just, for the artwork produced reflects their own unique ideas, experiences, and imagination. While it may be argued that ownership over AI art should reside elsewhere, this paper has proven that the users of GAI are the rightful owners. This paper emphasises that ownership does not belong to the GAI’s developers, the artists who inspired the GAI’s outputs, nor the GAI itself. Additionally, it explores the importance as to why the users of GAI should have ownership over GAI outputs, as they have a moral responsibility over the art they produce. Ultimately, the ownership over AI artwork should reside with the users of GAI, because this is the most righteous option available. By assigning ownership to these users, we can ensure that these users are rightfully held accountable for the work in which they produce, and uphold authorship over works which are representative of their ideas and creativity.
Appendix:
AI has been used to form and construct certain aspects of this paper. In this paper, I have exclusively used the LLM (Large Language Model) developed by Bing. Aspects of this paper which have been produced by this LLM have been highlighted in green. Additionally, aspects which have paraphrased the LLM’s outputs and responses have been highlighted in yellow. Specifically, the LLM was utilised to summarise Zurth’s argument(s), the prompts given to this LLM are as follows:
‘Summarise Zurth, Patrick. “Artificial Creativity? A Case Against Copyright Protection for AI-Generated Works.” UCLA Journal of Law & Technology (2021): 2-20. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3707651’
Followed by: ‘Why does Zurth conclude this?’
Bibliography:
Chen, Min. “A Scientist Has Filed Suit Against the U.S. Copyright Office, Arguing his A.I.- Generated Art Should Be Granted Protections.” News.artnet.com. 2023. https://news.artnet.com/art-world/ai-art-intellectual-property-lawsuit-stephen-thaler-2242031
References:
Cassel, David. “Stephen Thaler Claims He’s Built a Sentient AI.” Thenewstack.io.com. 2023. https://thenewstack.io/stephen-thaler-claims-hes-built-a-sentient-ai/.
Cetinic, Eva., She, James. “Understanding and Creating Art with AI: Review and Outlook.” Cornell University, (2021): 10. https://doi.org/10.48550/arXiv.2102.09109
Eliot, Lance. “Does Exposing and Logging those Generative AI ChatGPT Erroneous and Hallucinatory Outputs do Any Good, Asks AI Ethics and AI Law.” Forbes.com. 2023. https://www.forbes.com/sites/lanceeliot/2023/01/18/does-exposing-and-logging-those-generative-ai-chatgpt-erroneous-and-hallucinatory-outputs-do-any-good-asks-ai-ethics-and-ai-law/?sh=7ef1984c6004
Eliot, Lance. “Generative AI That’s Based on The Murky Devious Dark Web Might Ironically be The Best Thing Ever, Says AI Ethics and AI Law.” Forbes.com. 2023. https://www.forbes.com/sites/lanceeliot/2023/05/21/generative-ai-thats-based-on-the-murky-devious-dark-web-might-ironically-be-the-best-thing-ever-says-ai-ethics-and-ai-law/?sh=4fd5157e65b5
Hals, Toms., Brittain, Blake. “Insight: Humans vs. Machines: The Fight to Copyright AI Art.” Reuters.com. 2023. https://www.reuters.com/default/humans-vs-machines-fight-copyright-ai-art-2023-04-01/
Hayler, Kevin. “Can You Copy Art and Sell a Painting of a Painting? I Found Out.” Wildlifeartstore.com. https://www.wildlifeartstore.com/can-you-copy-art/#:~:text=It%20is%20legal%20to%20copy%20anything.%20It%20is,substantially%20similar%20to%20another%20original%20work%20of%20art.
Hertzmann, Aaron. “Who Should Get the Credit for AI Art?” edition.cnn.com. 2019. https://edition.cnn.com/style/article/ai-art-who-should-get-credit-conversation/index.html\
Investopedia. “Generative AI: How It Works, History, and Pros and Cons.” Investopedia.com. 2023. https://www.investopedia.com/generative-ai-7497939#:~:text=Generative%20AI%2C%20or%20generative%20artificial%20intelligence%2C%20is%20a,or%20images%20based%20on%20user-given%20prompts%20or%20dialogue.
Matulionyte, Rita. “AI as an Inventor: Has the Federal Court of Australia Erred in DABUS?” Journal of Intellectual Property, Information Technology and Electronic Commerce Law 13, no. 2 (2022): 1-20. https://www.jipitec.eu/issues/jipitec-13-2-2022/5538
McCormack, Jon., Gifford, Toby., and Hutchings, Patrick. Computational Intelligence in Music, Sound, Art and Design: Autonomy, Authenticity, Authorship and Intention in Computer Generated Art. Melbourne: Springer, 2019.
McFadden, Christopher. “The Rise of AI Art: What is it, and is it Really Art?” Interestingengineering.com. 2023. https://interestingengineering.com/culture/what-is-ai-generated-art
Miya, Mataichi. “What Is Art?” The North American Review 217, no. 811 (1923): 829. http://www.jstor.org/stable/25113044.
Zurth, Patrick. “Artificial Creativity? A Case Against Copyright Protection for AI-Generated Works.” UCLA Journal of Law & Technology (2021): 2-20. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3707651
[1] “Generative AI: How It Works, History, and Pros and Cons,” Investopedia, 2023, https://www.investopedia.com/generative-ai-7497939#:~:text=Generative%20AI%2C%20or%20generative%20artificial%20intelligence%2C%20is%20a,or%20images%20based%20on%20user-given%20prompts%20or%20dialogue.
[2] Mataichi Miya, “What Is Art?” The North American Review 217, no. 811 (1923): 829, http://www.jstor.org/stable/25113044.
[3] Eva Cetinic and James She, “Understanding and Creating Art with AI: Review and Outlook,” Cornell University, (2021): 10, https://doi.org/10.48550/arXiv.2102.09109.
[4] McCormack et al. Computational Intelligence in Music, Sound, Art and Design: Autonomy, Authenticity, Authorship and Intention in Computer Generated Art (Melbourne: Springer, 2019), 43-44.
[5] “Can You Copy Art and Sell a Painting of a Painting? I Found Out,” Wildlifeartstore.com, https://www.wildlifeartstore.com/can-you-copy-art/#:~:text=It%20is%20legal%20to%20copy%20anything.%20It%20is,substantially%20similar%20to%20another%20original%20work%20of%20art.
[6] “The Rise of AI Art: What is it, and is it Really Art?” Interesting Engineering, 2023, https://interestingengineering.com/culture/what-is-ai-generated-art.
[7] “Who Should Get the Credit for AI Art?” The Conversation, 2019, https://edition.cnn.com/style/article/ai-art-who-should-get-credit-conversation/index.html.
[8] “Insight: Humans vs. Machines: The Fight to Copyright AI Art,” Reuters, 2023, https://www.reuters.com/default/humans-vs-machines-fight-copyright-ai-art-2023-04-01/.
[9] “Stephen Thaler Claims He’s Built a Sentient AI,” The New Stack, 2023, https://thenewstack.io/stephen-thaler-claims-hes-built-a-sentient-ai/.
[10] “Does Exposing and Logging those Generative AI ChatGPT Erroneous and Hallucinatory Outputs do Any Good, Asks AI Ethics and AI Law,” Forbes, 2023, https://www.forbes.com/sites/lanceeliot/2023/01/18/does-exposing-and-logging-those-generative-ai-chatgpt-erroneous-and-hallucinatory-outputs-do-any-good-asks-ai-ethics-and-ai-law/?sh=7ef1984c6004.
[11] “Generative AI That’s Based on The Murky Devious Dark Web Might Ironically be The Best Thing Ever, Says AI Ethics and AI Law,” Forbes, 2023, https://www.forbes.com/sites/lanceeliot/2023/05/21/generative-ai-thats-based-on-the-murky-devious-dark-web-might-ironically-be-the-best-thing-ever-says-ai-ethics-and-ai-law/?sh=4fd5157e65b5.
[12] Rita Matulionyte, “AI as an Inventor: Has the Federal Court of Australia Erred in DABUS?” Journal of Intellectual Property, Information Technology and Electronic Commerce Law 13, no. 2 (2022): 1-20, https://www.jipitec.eu/issues/jipitec-13-2-2022/5538.
[13] Patrick Zurth, “Artificial Creativity? A Case Against Copyright Protection for AI-Generated Works,” UCLA Journal of Law & Technology (2021): 2-20, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3707651.
Comments