AI-Generated Fashion Designs: Ownership and Intellectual Property Challenges

Abstract

Fashion is being negatively impacted by artificial intelligence as it can produce clothing design(s), pattern(s), colour(s), and/or style(s) of fashion and/or designs to establish new fashion trends through machine learning (ML) programmes. These advances in technology promote Improved efficiency, creativity, etc., but they create numerous unique legal issues regarding intellectual property and ownership rights. In this article, the focus is on determining if an AI-created fashion design may be protected under the current forms of intellectual property (IP), specifically copyright and design law. An examination will be performed on the need for a human author of an AI-created work(s) under copyright law, along with an assessment of the ability of current IPR laws to adequately deal with works created through non-human creativity. A review of applicable legislative statutes, international policy activities and court opinions regarding the application of copyright to works created by non-humans will aid the writer to illustrate that there is legal uncertainty regarding all works created by artificial intelligence (AI). The conclusion is that the current IPR laws are based on the direct human relationship to IPR and therefore, are inadequate to deal with the effects of AI impacts to the apparel/fashion industry; thus, these laws require major legislative amendments and clarifications in order to adequately address the effects of AI on fashion design.

Introduction

Historically speaking, humans have been the source of creative inspiration in the creation of clothing. The use of creativity and artistry in making unique outfits has always been at the core of fashion. However, advancements in technology have allowed fashion designers to increasingly rely on AI technologies to produce innovative ideas for their clothing collections and to predict future trends based upon previous designs seen within the marketplace. Many fashion designers and technology companies now utilize machine learning algorithms in order to analyze vast amounts of data on previous designs they created, the buying preferences of consumers, and the patterns shown within the fashion marketplace. This information provides designers with a new way to create and visualize new garments.

While AI offers much promise from both a business and creative standpoint, there are also many issues that arise legally as a result of AI’s introduction into the field of fashion design. Intellectual property laws, such as copyright and design protection laws, traditionally consider creativity and authorship as being unique properties of humans. This thus raises questions about whether or not an output produced solely by an AI system (i.e. no human creator) would qualify for protection under existing intellectual property laws and, if so, who would receive rights to such output.

This problem is of particular import to fashion law since originality in design and brand identity are core components to the global fashion economy. In the absence of clear legal protections, designers and fashion business cannot adequately protect their own AI generated designs from being copied or commercially exploited by their competitors.

The major research question that this article will attempt to answer is: what kind of protection (i.e. copyright or design rights) can AI generated fashion designs receive, from an existing copyright design framework, as well as who, is the legal author or owner of these works?

To answer this question, the article will first provide an overview of the legal frameworks of copyright and design protection. The article will then examine how AI generated works present challenges to traditional doctrines of authorship through a critical analysis of relevant judicial decisions (i.e. courts) regarding creativity by non-human authors and how the courts have approached the problem of determining authorship. The article will conclude that current intellectual property laws (i.e. copyright and designs) are not designed to address AI generated fashion designs and thus will need to be reformed legally and politically.

Background

Machine intelligence is a general term to signify a computer process that demonstrates an ability to execute functions requiring human-like cognition such as ; recognizing patterns, making predictions, and creating new ideas through creativity and imagination. The newest generative AI platforms are able to analyze vast amounts of visual information (database) and then to provide outputs that create new original images that mirror human-created designs. Consequently, the fashion industry is increasingly using these platforms to support their designers by helping them visualize their designs, generate textile printing designs and help forecast future design trends.

From a legal perspective; fashion designers can typically protect their designs based on a combination of laws, particularly copyright, design and trademark laws. Copyright law protects artistic works that can include sketches/sketch drawings/graphic designs, etc., which are all utilized in fashion. In India, the Copyright Act (1957) provides copyright protection for artistic works including drawings and any design that can be represented in a tangible form.[1] However, copyright protection usually requires that the original artwork is created by an actual human author, and that the author used skill, through the use of human judgement, and incorporated labor in developing the original artistic product.

Like fashion trademark law, a design protection regime allows for the protection of the aesthetics of industrial manufactures to be applied to manufactured products. The Designs Act 2000 allows a manufacturer of new and original designs applied to articles (which include patterns, shapes and ornamental features of a garment or accessory) the ability to apply for legal protection of their design.[2] The focus of design protection is on the visual appearance of a design rather than a designer’s artistic authorship. Furthermore, a design will only be permitted to be registered with a design if an applicant for registration of the design is a legal entity entitled to claim ownership of the subject design.

At the global level, there are also international agreements that provide intellectual property protection for creative works, such as the Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPs), which sets the minimum standard for copyright and design protection among the TRIPS member states.[3] Many organisations are now exploring the use of artificial intelligence in relation to the legal implications for intellectual property law. In March 2022, the World Intellectual Property Organisation conducted various consultations to better understand who owns AI-generated works and how such works may be protected – leading it to a conclusion that there continues to be a significant amount of uncertainty regarding authorship and ownership of AI-generated works in the intellectual property context.[4]

The heart of the confusion involves how intellectual property law has always focused on how works are created by humans. That creates an issue when you have products made by artificial intelligence systems with very little human input, as this raises questions of whether or not these products will be seen to have met the standards of originality and authorship necessary under traditional IP laws in order to be awarded IP protection (i.e., copyright).

Legal Analysis

As fashion continues to evolve and utilize artificial intelligence (AI), some of the most fundamental elements of the law will need to be reevaluated in order to determine what level of protection should be afforded, if any, to these intangible creations born from artificial intelligence (AIs). One of the main areas of concern is that under the current IP regime, all intellectual property protection (e.g., copyright) is dependent upon authorship being human (since copyright was created for human authors). This view of at authorship (as being distinctly human) has been reflected in the laws of various jurisdictions, as they have established certain legal principles/configurations that define authorship as an activity that requires human creativity, intellectual effort and independent judgement to produce a work that is entitled to protection. Thus, because AI does not provide any measurable level of creativity independent judgement and/or humanly engaging any of those elements while creating their works (e.g., generating designs, using software applications, etc.), the works are then unable to be entitled to copyright even if they otherwise satisfy the criteria established through existing law (i.e., originality).

It has been suggested that a human author can be attributed to the AI-generated work, depending on how much creative control they exercised over the final output, such as if a fashion designer used an AI program to generate fashion designs but maintained sufficient creative control over the final output. However, this theory creates issues if an AI system generates design variations independently of any direct input from a human designer; there would not be an identifiable individual associated with the creative expression embodied in the design.

Another argument for ownership of AI-generated work is the developer or company that created the AI program or algorithm. The developer designed the algorithm, which ultimately produced the final product. Therefore, the output should legally belong to the developer. However, this argument also has its own set of issues because software developers do not typically dictate specific creative outputs generated by the AI program once the program has been made available for use by consumers or clients.

One of the greater assumptions made about the copyright status of AI-generated works is that they should not have any copyright protection, but fall into the public domain. Legal experts have expressed opinions that granting IP rights to machine-created works will diminish the purpose of copyright law as a way to promote human creativity; since there is no creativity involved on the part of the human, there may not be a reason to grant exclusive rights to either the author or AI.

In relation to the fashion industry, these issues could create significant problems for fashion companies who have to put a lot of time and money into creating their designs and developing their brand. If protection for AI-created designs cannot be secured under design or copyright laws, designers may have a hard time enforcing their intellectual property rights against imitators and/or unauthorized reproductions by competitors.

Design laws may provide some degree of protection for AI-created designs; however, design protection primarily looks at the appearance of the design and does not account who created the design. The issue, still, remains that many design laws will only provide protection for the design if it is owned by a human being; thus, AI-created designs leave the question of who owns the design in limbo.

The legal frameworks that govern intellectual property rights have yet to evolve adequately for the challenges that artificial intelligence (AI) poses to fashion design. In the meantime, without clear legal rules regarding authorship and ownership of AI-created works, designers, fashion brands, and developers of technology will remain uncertain as to what their right of ownership is and how it will affect their decisions in doing business in the future.

Case Analysis

The few courts that have examined the issue of non-human authorship (i.e., where animals or machines create works) have provided valuable insights into how courts will deal with issues relating to intellectual property and non-human authors. To date, while there have not been specific cases regarding works created through artificial intelligence in fashion design, there are a number of cases which have set precedent for general legal principles any case involving a work created by AI would rely on in determining whether or not AI-created works are protected under copyright law.

One relevant case is a decision made by the Ninth Circuit Court of Appeals, Naruto, et al. v. David Slater.[5] In this case, a wildlife photographer had set up a camera which was then taken by a crested macaque monkey and the monkey subsequently recorded a photograph of itself. The plaintiff’s contention was that the monkey was entitled to copyright protection for the photograph that it had taken of itself. However, the court found that the monkey was not the author of the photograph and therefore, under the United States copyright laws, had no right to copyright protection for the photograph it had taken of itself. The court relied upon the fact that the United States copyright laws limit copyright protection to human authors to determine that since copyright has historically been limited to human authors, that in addition, through both statutory interpretation and through a long history of case law, only authors who are human would be afforded copyright protection.

This ruling on copyright law as it relates to AI-generated works is also significant. Copyright may not apply to entirely AI-generated works (as no human was an author of these works because they were written entirely by one or more AIs) and since the requirement for copyright protection is a human-created work, there are no grounds under which such a work would likely be protected by copyright.

The UK Supreme Court (in a case that raises similar questions concerning the rights of AI inventors under patent law) recently issued its decision in Thaler v. Comptroller General of Patents, Designs and Trade Marks regarding the patentability of inventions created by an AI named ‘DABUS’.[6] The Supreme Court concluded that to qualify as an inventor under the Patents Act, a person must be a natural person, so therefore DABUS, could not legally be an inventor.

Although the Thaler case was a decision on patent law and not copyright law, it reinforces the concept that IP laws exist to protect and promote the interests of creators of works of authorship, so that there is an incentive to create, and there is a financial benefit associated with this creation. The reasoning in this decision must be viewed in light of the fact that Canadian law also provides that an inventor must be a natural person. This creates an assumption and legislative intent that inventors (and by analogy authors) must be natural persons, and therefore capable of having both legal liability and rights.

Also, in Acohs v Ucorp, the Federal Court of Australia addressed whether copyright can be granted for material produced by a machine or computer, concluding that human authorship must be involved to receive copyright protection and that a computer-generated work cannot have copyright due to the lack of human creativity in the computer-generated work.[7]

The courts have demonstrated a consistent analytical framework, and the doctrine of intellectual property has remained consistent across all jurisdictions and is largely dependent on the human authorship of the work to be granted intellectual property protection. Therefore, works solely generated by machines may face difficulties in securing rights within the parameters of existing laws.

Critical Analysis and Findings

The above discussion provides evidence that artificial intelligence is creating significant challenges to the foundations of intellectual property law, particularly regarding the legal concept of authorship required to receive copyright protection; however, as generative artificial intelligence continues to evolve, its outputs will become nearly indistinguishable from human beings.

In the area of fashion, this distinction between the output of generative artificial intelligence and that of a human creator will continue to be addressed as fashion brands embrace generative artificial intelligence as part of their design process, to obtain predictions for fashion trends and generate visual prototypes. Uncertainty will continue to persist until there are definitive legal industrial rules regarding the ownership of rights in relation to the designs produced by generative artificial intelligence.

One potential response to this concern is to treat the human user of an AI system as the legal author of the work produced by that AI system, provided that the user has exercised meaningful creative control over the output. This approach to copyright law maintains a structure that is primarily human-centred while also recognising the collaborative role played by AI tools in the creative process.

Alternatively, legislators could create a new category of IP protection specifically for works generated by AI. Under this type of system, ownership of AI-generated works may rest with the entity that operated the AI system and/or the entity that funded the creation of the work. The creation of new rights may lead to the unnecessary expansion of IP protection and/or the possible negative impact on competition and innovation.

The ideal solution will most likely involve a combination of doctrinal clarification and policy change. Courts and legislatures should clarify the amount of human creativity necessary to establish authorship in a work produced with the assistance of AI. Without this type of clarification, the legal status of fashion designs generated by AI will remain unknown, which will ultimately impact whether or not there is innovation in and/or sufficient protection for fashion industries.

Conclusion

AI-generated creativity has had an impact on the way we create fashion design and therefore, will continue to shape the evolution of the fashion industry as emerging technology allows for innovative forms of fashion design to be generated by more sophisticated computing systems. Despite these technological advancements, current intellectual property regimes operate principally under the premise that works of creativity are created by human authors.

This paper examines various legal issues regarding AI-generated fashion designs within copyright and design law frameworks. A review of the relevant legislation, international policy, and court decisions concerning the issue of authorship of non-human beings indicates that the existing intellectual property system is ill-equipped to address machine-generated creativity. Courts have consistently highlighted the requirement of human authorship, which limits the possibility of obtaining protection for works generated autonomously by artificial intelligence.

In order for the fashion industry to promote innovation and provide legal certainty, policymakers must assess how intellectual property law can respond to creativity generated by artificial intelligence. It will be necessary to clarify standards of authorship and to create an appropriate regulatory framework with respect to the objectives of protecting intellectual property and promoting technological innovation.

Reference(S):

[1] The Copyright Act 1957.

[2] The Designs Act 2000.

[3] Agreement on Trade-related Aspects of Intellectual Property Rights 1994.

[4] World Intellectual Property Organization, WIPO Conversation on Intellectual Property and Artificial Intelligence (WIPO 2020).

[5] Naruto, et al. v. David Slater (United States Court of Appeals)

[6] Thaler v. Comptroller General of Patents, Designs and Trade Marks (Supreme Court of the UK)

[7] Acohs Pty Ltd v Ucorp Pty Ltd (Federal Court of Appeal, Australia).