[This article is authored by Tejaswini Kaushal, a 2nd year B.A. LL.B. (Hons.) student at Dr. Ram Manohar Lohiya National Law University, Lucknow. It elucidates the international and national status quo of AI-generated art and reflect on the ethical and legal standards. This is part of a series titled ‘AI Art,’ the second article of which can be found here.]
In December 2022, Ammaar Reshi‘s children’s book “Alice and Sparkle” was an instant hit among the children and a money-puller for Ammaar, especially considering how he curated this 12-page picture book in a few days. Yet, several artists were not happy with his creation. These were those artists whose works of art had been modified and presented for use by an AI tool, ChatGPT, which Ammaar had utilised to curate the content. This instance of the use of AI has caused professional illustrators and authors to fear that AI-generated art and stories may threaten their livelihoods. Critics have argued that AI-generated works are not of human authorship and lack the originality, creativity, and imagination of human-made works. In addition, some have argued that the AI models used to generate the works are trained on other people’s work, thus stealing from other human artists.
It is evident that though AI acts as a pivotal utility tool for most creators, especially in this age of digital art and NFTs, art created by them poses equally significant ethical challenges regarding copyright infringement and ownership rights of works created. This article will elucidate the international and national status quo of AI-generated art and reflect on the ethical and legal standards.
AI-generated Art Trends around the Globe
Artificial intelligence (AI) has become a growing trend in the art world, as more and more artists are turning to AI tools to produce artwork. Some recent auctions demonstrate the increasing demand and value for AI-generated art enjoys in today’s tech-savvy world order. In October 2018, a painting generated by an artificial intelligence system called “Obvious” sold for $432,000 at an auction, the highest for a piece of AI-generated art. This instance was the first to raise questions about whether AI-generated art should be treated similarly to human-generated art and whether the algorithm should be credited as the author of the work. Another notable example is the case of AI artist Mario Klingemann, whose paintings were the first AI-generated artworks to be exhibited in a significant museum and led to questions about whether AI-generated art should be considered “real” art. With well-known publications like The Economist and Cosmopolitan utilising AI-generated artwork for the front covers of their June 2022 issues, artificial intelligence (AI)-generated artwork has already begun to be used in popular media. Furthermore, in September 2022, the 300 USD prize money from the annual art competition at the Colorado State Fair has also been won by the use of an AI graphic-generating tool, creating the work named “Théâtre D’opéra Spatial”.
Indubitably, AI-generated art has become increasingly popular in recent years, with many galleries and museums around the world showcasing AI-generated artworks. AI tools, such as DALL-E 2, NightCafe, Artbreeder, StarryAI, etc., have also been used to create interactive experiences, such as virtual reality installations, interactive sculptures, and augmented reality art. AI tools have also been used to create unique paintings, sculptures, and varied works of art, from abstract expressionist art to photorealistic oil paintings. The use of AI has allowed for the creation of art that is more intricate and complex than ever before. AI-generated art has also been used to create thoroughly unique works that can be manipulated and modified by the user, allowing for a whole new level of interactivity and creativity.
How does AI create Art?
Artificial intelligence art is created through the use of algorithms and computer-generated images. AI-generated art generators create unique pieces of artwork by taking an initial mapping of a text description and then stochastically generating from it. In order to make AI-generated art, artists design algorithms that “learn” a particular aesthetic by examining a large number of photos rather than following a set of rules. The program then strives to produce fresh images while adhering to the learnt aesthetics.
A class of algorithms known as Generative Adversarial Networks (GANs) has been employed in the majority of AI-generated artworks that have surfaced in recent years. With a GAN, two sub-models are trained at the same time. The first is a generator, which creates random images, and the second is a discriminator, which is trained to distinguish between real and fake photos. GANs operate according to the game theory, and they are deemed successful when they produce a fake specimen that is sufficiently believable to mislead both a discriminator and humans.
Ethical concerns regarding AI-generated art
The lack of privacy and consent from artists whose work is used to train AI models, the potential for AI-generated art to replace human art and its potential to be used for commercial gain without compensating the artists are certain concerns plaguing this arena. It also raises questions about who should be credited as the author of the work and whether AI models should have the same copyright protections as human artists. In addition, it probes the potential for AI-generated art to eternalise detrimental stereotypes and be used to create sexualised images, like in the instance of artist Robbie Barrat, who used an AI algorithm to create a series of nude portraits. Finally, it raises questions about the potential for AI to be used to manipulate public opinion and spread misinformation. Even Ammaar, to allay the concerns raised against his usage of copyrighted images, has advocated for greater transparency from AI platforms regarding their data sources, as well as for platforms to work with artists to ensure that their rights are protected.
Laws governing AI-generated Art in India
Currently, there are no specific laws in India that relate to Artificial Intelligence (AI), Big Data (BD) or Machine Learning (ML). The Government’s priority at this stage seems to be in the promotion of AI and its usage. However, the Copyright Act, 1957 does provide some protection in the case of computer-generated works under Section 14(b) of the Act. The Act states that copyright is granted to the person who causes the work to be created. In the case of literary, dramatic musical or artistic creation, which is computer generated, the author is the person who causes the work to be created. It is still debatable if AI can be named as an author of a piece. There is no Indian precedent to suggest that a computer can be given authorship. The Indian Copyright Office similarly provides no policy guidance on whether only humans can be deemed writers. Yet, AIs and computer software will be liable for indulging in copyright infringement at par with natural persons.
In 2021 the Indian Copyright Office granted an AI, RAGHAV (Robust Artificial Intelligent Graphics and Art Visualizer), the status of co-author for a piece of artwork, which it later tried retracting unsuccessfully. However, it is unclear what the basis for this decision was and what is the current position on the case since the copyright remains ‘registered’. Furthermore, in 2021, the Parliamentary Standing Committee Report also recommended revising the current Patents Act and Copyright Act to introduce AI and inventions related to AI so as to “extract benefits from AI” in recognition of the increasing relevance of AI in multiple disciplines as well as its economic impact.
The International Standpoint of Copyright Protection to AI-generated art
In the US, copyright laws regarding AI-generated art are still developing. Currently, the US Copyright Office does not grant copyright registrations for AI-generated works, as these works are not considered to be of human authorship. However, there are some cases where a human author has been credited with the work, and the Copyright Office has granted copyright registrations for these works. In November 2022, the US Supreme Court held in the case of Thaler v. Vidal that an artificial intelligence program (named “DABUS” in the present case) could not claim a patent on an invention containing no traces of “human authorship”.
However, some jurisdictions do have flexible copyright law regimes, which have provided protections for AI-created works. For instance, unlike in the US, South Africa was the very first nation to grant a patent as the ‘author’ to DABUS in July 2021. Furthermore, The Australian Copyright Act provides protection for “original literary, dramatic, musical or artistic works” and does not specify that these works must be created by a human. In light of this, DABUS was held to be an inventor under the Australian patent regime as well.
In France, as per Article L. 112-1 of the French Intellectual Property Code, only original works by a natural person as its author can be copyrighted, leaving computers and soft wares beyond the Code’s purview. In Canada, the Canadian Copyright Act, 1985 fails to cover AI-generated content under its ambit and also has no legal precedent on this matter. In the UK, copyright law grants authorship rights to the creator of the AI system under Section 9(3) of the Copyright Designs and Patents Act, 1988. In our neighbouring country of China, two court decisions of 2019 have set the main base for this discussion. In the Shenzhen Tencent v. Shanghai Yingxun case, the copyright for AI inventions was recognised, although it was refused in Feilin v. Baidu on the contention of lack of originality as not a human creation. Both cases recognised the work as that of a legal entity, creating a global expectation of China’s acceptance of AI creators in the near future.
Guidelines by WIPO
The World Intellectual Property Organization (WIPO) has stated that owners of AI systems should have the right to patent inventions generated by their AI systems. Under WIPO’s guidelines, the owner of an AI system should receive patent protection for AI-generated inventions as long as the invention meets the criteria for patentability. WIPO also states that patent law should be flexible enough to allow AI systems to be named as the inventor of a patent as long as the AI system meets the criteria for inventorship. In addition, WIPO has recommended that governments review their patent laws to ensure that AI systems can be included in the patent filing process.
Artificial intelligence (AI) is set to reshape the art world in the coming years and is already making a significant impact. AI-generated art is a nascent trend that has emerged due to the development and exponential growth of AI and machine learning technologies. AI artists promote the use of AI as a superior alternative to human beings when it comes to artistic creation. As a novel algorithm-run phenomenon with no established laws that govern it, it may infringe the copyrights of other artists. AI-generated art holds the potential to result in an artistless world, where AI is the sole determinant of what is considered art and what is not. Additionally, AI-generated art can have a potentially negative impact on the livelihood of human artists, who may lose their jobs to AI models. In order to tackle these concerns, it is crucial that the copyright standards must evaluate this form of art.