This article is written by Sneha Singhal, Galgotias university, BALLB(Hons) 4th year student during her internship with LeDroit India.
Abstract
AI’s rapid advancements have modernized many industries, including the artistic domain. Advanced AI, especially with deep learning and generative capabilities, can now independently produce complex works with minimal human input. Yet, two significant legal hurdles stick around i.e. the copyrightability of AI generated works remains a contentious issue, and establishing liability for such content poses substantial Challenges for legal jurisprudence. Consequently, there’s now a considerable legal gap in how AI-generated work is protected, owned, and regulated.
This article explores how AI generated originality impacts copyright regulations. It critically examines the initial challenges AI content poses to existing copyright laws, reviewing current legal frameworks and relevant case laws. The article emphasizes the urgent need for legislative updates to address the challenges and opportunities presented by AI- generated works.
Keywords: Artificial intelligence (AI), Copyrightability, AI-generated work, Ownership, Legal gap,
Introduction
Artificial intelligence (AI) has truly seized the cultural imagination and is quickly becoming central to how we live. At its core, AI, especially its generative forms, promises greater efficiency. People started not using their minds to write or create a single thing. Need a quick chat? ChatGPT’s there. Want to create an image? Midjourney’s got you covered. Truly, AI is built for every academic discipline, making users excited to explore its capabilities.
AI, while designed to ease burdens and assist those with heavy workloads, is increasingly being misused. Instead of serving as a helpful tool, it’s often exploited for shortcuts and convenience, diminishing its intended beneficial purpose for human users.
The realm of copyright law, built to protect the profound investment of time, effort, brain, skills in creative endeavours, is now confronted by artificial intelligence’s ability to instantly produce the work, threatening the very essence of human originality that once took months to cultivate.
The core question revolves around whether AI generated work deserves copyright protection. If they do, how will this be regulated in India, given the current legal landscape. Is AI trying to OutSpace human imagination by bringing cheap creativity into the race? Presently, India Information technology laws provide provisions related to AI and it doesn’t have its own regulatory legislation. Meanwhile, the copyright act of 1957 traditionally protects works created by human authors. This creates a legal vacuum for AI generated works. The challenges lie in determining how to protect copyright for AI generated content within the existing legal framework, or if new legislation is needed to bridge the gap.
The purpose of this study is to ascertain the longevity and global protectability of AI generated works, evaluating their preparedness for future adoption. It will delineate the requisite safeguards for virtual creations in the context of industry 5.0, concurrently addressing the complex issue of authorship and ownership for AI generated content. The paper will conclude by proposing solutions and recommendations to mitigate existing challenges.
Understanding AI generated content
AI generated content encompasses various forms of media, including text, video, code, and audio, all produced by generative AI tools. These tools are trained on extensive datasets, enabling them to generate relevant output in response to diverse inputs such as words, phrases or questions.
The core distinction between AI assisted creation and AI generated works lies in human involvement. In AI assisted creation, a human author utilizes AI as an aid, maintaining final creative authority and copyright. In contrast, AI generated works are largely or entirely produced by the AI system, requiring little to no human input.
The concept of AI generated work, where the entirety of the creation is attributed to artificial intelligence and a human prompt-giver subsequently claims questions. We can’t fully grasp AI generated works without also looking at the idea of originality, a common prerequisite for copyright in many jurisdictions. Human creativity and intellectual effort are hallmarks of traditional originality. AI creates content by using algorithms to analyse and reassemble existing information. To address the legitimacy of such claims, it is crucial to determine whether AI generated works are eligible for copyright protection, to understand the historical development of this issue and to identify which countries currently offer legal protection for AI generated content.
Can AI generated work be given copyright protection?
The long-debated question of copyright protection for AI generated content often yields a “NO” in many jurisdictions. Yet, the landscape isn’t uniform, some jurisdictions are beginning to grant such protection. This evolving situation leads to a key query: if AI generated work does become copyrightable, who would be the right claimant?
Let’s delve into the potential candidates individually.
- ChatGPT or other AI platforms
ChatGPT cannot be an author under the copyright act, 1957 as the definition of author is the person who creates it. The term “person” generally means a natural human or a legal person which includes company, association, firms and others. The outputs of ChatGPT are generally not eligible for copyright protection themselves.
- AI developer as author?
Whether developers hold the copyright to AI generated work largely hinges on their terms of use policy. If their Term of use states they retain rights to output from their AI, then they typically own the copyright, unless a separate contract specifies otherwise. However, the terms of use for most popular Ais., like chatbots, generally don’t claim copyright over the content. It is also not right in the legal realm as OpenAI is only a service provider and anything derived from that should not give an automated protection to the developer as it depends upon the person how they choose to use the service. It would not be right to say that a developer can be an author or should be given an absolute right over the AI generated work.
- Can a prompt giver be granted protection?
The Indian copyright act was amended in 1994 to address the authorship of computer-generated artistic works. This amendment introduced an additional provision, section 2(d)(vi) explicitly stating that the authorship of such works belongs to the person who caused their creation. It would be incorrect to assume that the phrase “causing the work to be granted” clarifies the position of law in India regarding copyright. A single line is insufficient for copyright protection, instead, an author’s artistic work must meet the requisite standards of originality to qualify for protection under copyright law.
The lowest standard for protection known as the sweat of the brow doctrine, asserts that an author can secure copyright simply by demonstrating the effort and diligence invested in creating their work. This doctrine prioritizes the labour expended rather than requiring any significant creativity or originality in the work itself.
University of London Press v. University Tutorial Press, in this case, the university of London sued a publisher who compiled its past exam papers, claiming copyright infringement on behalf of the professors who authored them. The publisher countered that the papers lacked originality, delivering from an existing body of knowledge, and thus were not subject to copyright. The court held that, even if a paper draws from existing knowledge, the effort and diligence put into its creation are sufficient for copyright protection. These standards grant copyright based on the work expended, without requiring additional creativity.
Test of originality in copyright
The term original mentioned in Section 13(1)(a) of The Copyright Act, 1957 means as per the case of Rupendra Kashyap v Jiwan Publishing House, 1996, the term “original” does not refer to novelty of the ideas themselves. Instead, it signifies that the work must not be a copy of someone else’s creation. It must, therefore, originate from the author’s own labour and skill. In essence, “original” in this context means that the work has been independently created by the author and is not duplicated from another source.
Once tied to the “sweat of the brow”, the concept of originality now often requires only a modicum of creativity. However, since there’s no specific meaning, courts and authorities worldwide have varying criteria for determining originality, leading to diverse interpretations across countries.
Legal framework: current global landscape
India’s copyright act of 1957 currently limits copyright protection to works created by humans, whether a living individual or a legal entity. The act does not explicitly or implicitly recognize AI authorship, meaning that AI generated art is not eligible for copyright protection under the current law.
Section 2(d) of the act, describes various kinds of authors for each subject matter of copyright law. Clause (vi) herein provides author in relation to any literary, dramatic, musical or artistic work which is computer-generated, the person who causes the work to be created. This section offers limited recognition for authors of computer-based works, generally acknowledging the individual who causes the work to be created. Interpreting this, it suggests that the person providing the prompts would be considered the author.
The RAGHAV case: in 2020, the Indian copyright office encountered a unique situation when an AI system named RAGHAV attempted to register copyright for its network, “Suryast”. The initial application was rejected because a human author wasn’t identified. However, the painting was later granted protection after a natural person was listed as a co-author alongside RAGHAV. This led to a withdrawal notice being issued, seeking clarification on RAGHAV’s legal standing and highlighting the ambiguity surrounding whether an AI qualifies as an artist under the Copyright Act, 1957.
US copyright law mandates human authorship for protection. Works created without human intervention, like autonomously generated AI text or code are not copyrightable.
The case of Thaler v. Perlmutter, provides a detailed explanation to AI generated art and copyright, where Stephen thaler, an inventor, attempted to copyright a work produced by his generative AI system, “creative machine”. Thaler sought to name the AI as the author and then transfer the copyright to himself as the machine’s owner. However, the copyright office rejected his application. Following this, Thaler sued Shira Perlmutter. The registrar of copyrights, where the court held that AI generated artwork is not eligible for copyright protection if AI is the sole creator and there’s no human involvement in the creation process. This decision was based on the requirement of human authorship for copyrightability.
US law allows copyright for AI assisted works if a human contributes sufficient creativity, like creatively selecting and arranging AI outputs. However, purely machine-generated content is unprotected.
Under section 9(3) of the copyright, designs and patents act, 1988, UK law provides a degree of flexibility regarding the authorship of computer-generated works. It defines the author as the individual responsible for the arrangement leading to the word creation. This framework offers a mechanism for acknowledging AI generated work by assigning authorship to the human or entity overseeing the Ais functioning or design.
The European Union’s 2019 copyright in the digital single market directive sought to update copyright law for the digital age, but it hasn’t established clear rules for AI generated content. The EU intellectual property office and legal experts are still discussing how to address this issue. Current proposals suggest either expanding existing copyright laws to include AI or creating a new, specific legal framework called sui generis regime just for AI created works.
Legal and Ethical Challenges
- Legal liability and infringement- since, AI is not legally recognized as a person, determining civil and criminal liability for its harmful outputs such as defamation or incitement to violence remains a significant challenge. This complexity extends to copyright infringement by AI, as the absence of specific AI and copyright laws make it difficult to assign responsibility. Consequently, such problematic content might be removed from online databases, and in severe cases, the AI platform itself could face long-term bans.
- Authorship- Granting authorship to AI necessitates establishing it as a legal person. This would bestow upon AI rights akin to those of living individuals or other legal entities. However, this poses significant challenges in controlling Ais’s capabilities. Bestowing a human-like identity on AI would enable it to enter into contracts and more perilously, grant it the power to sue or to be sued. Furthermore, many countries grant moral rights to authors, which are intrinsically linked to human feelings and emotions which is a concept not applicable to AI.
- Lacking uniqueness and creativity- humans are mortal and produce a limited number of works based on their unique ideas and artistic skills, making copyright protection within a set time period straightforward. In contrast, AI is immortal, capable of rapidly generating countless works from its vast database, which could lead to repetitive art and make assigning a fixed protection period confusing and contentious.
- Originality- assessing the originality of AI generated content, especially when the AI pulls information from various published sources, regardless of their trustworthiness. This reliance on existing data, particularly unverified sources, often compromises the fundamental need for genuinely original work.
- Ownership and economic right- since there is no existing provision for AI to be recognized as an author, determining ownership of AI generated content, particularly the monetary benefits, becomes a significant challenge. This raises crucial questions about who receives income from protected AI generated content, leading to complex issues surrounding royalties and licensing.
There are many legal as well as ethical challenges associated with the copyright protection given to AI generated work. If AI generated work is to receive copyright protection, it demands a fundamental shift in legal frameworks. This includes substantial amendments to existing definitions and concepts, leading inevitably to changes in constitutional intellectual protection provisions and the recognition of AI as a legal entity.
Conclusion and Recommendation
Ais rapid advancement is quickly expanding its reach into areas of law traditionally requiring human skills and creativity. Given that AI, a human creation, now processes vast global data, it’s becoming increasingly powerful. This presents a complex challenge for copyright protection of AI generated works, necessitating significant amendments to domestic and international laws. While numerous international treaties and conventions safeguard creative works resulting from human skill and effort, the UKs CDPA, 1988 is currently unique in granting copyright protection to AI generated works. Countries like the US, China, EU, and India do not yet recognize AI generated content as original, leaving it in the public domain.
When it comes to securing copyright protection for AI generated works, a key solution involves granting shared authorship to both the individuals providing prompts and the Ais developer. This approach effectively resolves debates around legal liability in case of infringement. Furthermore, it addresses the originality question if prompt givers can modify AI generated content, even by incorporating elements from protected works, the resulting work could qualify under an updated definition of originality. This dual authorship also simplifies the process of assigning a protection period. To maintain uniqueness and creativity, an AI platform could be designed to restrict the total number of works an AI can create during its operational lifespan
Overcoming existing hurdles necessitates a robust AI law, crucial not only for intellectual property rights but for all other sectors. A dedicated legal framework for AI would unlock its full potential across divergent legal and non-legal landscapes.
References
