ABSTRACT
This article delves into the extensive array of intellectual property factors associated with AI-generated creations, concentrating on the subtle complexities of copyright law. The growing influence of Artificial Intelligence across various fields, including medical science, transportation, aviation, space exploration, education, entertainment (encompassing music, art, games, and films), and industry, has significantly altered our daily experiences. The domain of Intellectual Property Rights (IPR) is no exception to this transformation. The impact of AI on creativity and innovation has gained global acknowledgment, particularly in the realm of copyrights among the different categories of IPR. AI is capable of performing tasks such as composing music, writing blogs, novels, and poetry, as well as generating paintings and drawings. It is essential, however, to differentiate between works produced by individuals with AI assistance and those created solely by AI without human involvement. AI has introduced substantial challenges and raised critical issues within the sphere of intellectual property rights, especially concerning copyright law. This article examines the significance of AI in the creation of artistic works, including art, music, poetry, and novels. It also explores the legal frameworks in various countries and provides an in-depth analysis of different authorship models for AI-generated creations. Furthermore, the article addresses ongoing discussions at the World Intellectual Property Organization (WIPO) and examines the legal implications and consequences in India.
Keywords: Artificial Intelligence, Creation, Copyright, Author, Intellectual Property
INTRODUCTION
“Artificial intelligence is not a substitute for human intelligence; it is a tool to amplify human creativity and ingenuity.” – Fei-Fei Lee
ARTIFICIAL INTELLIGENCE (AI) has gained considerable significance in modern society, becoming essential in a wide array of technological applications. AI is a branch of computer science focused on the study and development of intelligent machines capable of thinking, learning, and problem-solving in ways akin to human beings. The objectives of AI encompass enhanced learning, reasoning, and perception through computational means. AI systems are adept at processing information, recognizing patterns, and making independent decisions. The advancements in data science and computing have streamlined the decision-making process, with applications spanning various sectors, including finance and healthcare. Deep learning, a subset of AI, enables computers to analyse data in a manner inspired by the human brain. The expanding influence of AI in creativity and innovation has garnered global recognition.
Today, individuals from all walks of life are acquainted with ChatGPT, a generative AI chatbot created by OpenAI, a research laboratory in the United States. This chatbot underwent extensive training, analysing thousands of digital books, the entirety of Wikipedia, and nearly a trillion words from blogs, social media, and other online sources to master natural language. ChatGPT is capable of composing poetry, generating tweets, answering trivia questions, summarizing emails, translating languages, and even writing its own computer programs. It demonstrates an understanding of the complexities of human language and can perform various human-like tasks. Moreover, AI is also capable of writing local news articles, creating artwork, composing short novels, and generating music by analysing different recordings. However, the rise of AI has raised significant concerns and challenges in the realm of copyright law. This article explores the importance of AI in the creation of artistic works, including art, music, and poetry, among other forms.
RESEARCH METHODOLOGY
This paper presents a descriptive analysis, utilizing secondary sources to explore the intersection of copyright law and artificial intelligence, along with the various legal implications and challenges that arise. The research incorporates information from secondary sources, including newspaper articles, academic journals, and dynamic websites.
REVIEW OF LITERATURE
The phrase “Artificial Intelligence” was first introduced by John McCarthy in 1956. Currently, there is no formal legal definition for “artificial intelligence.” It can be described as “the capability of machines to perform tasks that humans would typically associate with intelligence.” In 1990, Ray Kurzweil characterized AI as “the discipline focused on enabling computers to execute tasks that require human-like intelligence.”
Artificial intelligence generally pertains to the capacity of machines to engage in cognitive functions such as thinking, perceiving, learning, problem-solving, and making decisions. The World Intellectual Property Organization (WIPO) has identified three main categories of AI systems:
- “expert (or knowledge-based) systems”;
- “perception systems”; and
- “natural language systems.”
AI enables machines to perform tasks autonomously or with minimal human oversight, which would otherwise necessitate human intelligence. It should be understood not as a singular technology but as a broad field encompassing various subfields, including “machine learning, robotics, language processing, and deep learning.” Consequently, “machine learning” and “deep learning” are recognized as two subsets of AI. In the context of machine learning, an embedded algorithm within the computer program allows it to learn from data inputs, adapt, and make future decisions either independently or under guidance. Essentially, machine learning algorithms utilize the inputs provided by programmers to create new outputs by making autonomous decisions.
AI can facilitate the creation of two types of creative works: (i) “AI-generated” works and (ii) “AI-assisted” works. AI-generated works, also referred to as “autonomously generated by AI,” involve the production of content by AI without any human involvement. In this category, AI may adjust its behaviour during operation in response to unforeseen information or events, resulting in outputs that were not originally intended or anticipated. On the other hand, “AI-assisted” works involve human input alongside AI capabilities.
THE INTERSECTION OF COPYRIGHT LAW AND ARTIFICIAL INTELLIGENCE: ISSUES AND CHALLENGES FACED
Since the late 1900s, computer programs have been widely utilized in the creation of copyrighted works. Initially, these computer-generated creations did not pose significant challenges regarding copyright ownership. This was primarily because computer programs were viewed as mere tools that facilitated creative activities, necessitating human involvement for the actual production of the work. They were akin to stationery items that required human operators to generate creative outputs. However, the landscape has dramatically shifted with the advent of artificial intelligence (AI). Today, computer programs are no longer just tools; they possess the capability to independently generate works by making autonomous decisions. AI can produce a vast array of works with minimal investment and in a remarkably short timeframe. The creations generated by AI may be eligible for copyright protection across various jurisdictions due to their originality. The criterion of “skill and judgment” in determining originality may be considered fulfilled based on the “programming and parameters that guide the AI in compiling and creating the work.” Nonetheless, AI-generated works lack a designated author. In contrast, AI-assisted works involve human intervention, allowing the individual who facilitated the creation through the use of artificial intelligence to claim authorship. This distinction does not apply to works produced solely by AI without any human input. The question of authorship in such instances has become a complex issue that has perplexed nations worldwide.
Copyright protection acts as a motivation for authors to create additional works by utilizing their skills, effort, and judgment. If artificial intelligence (AI) is acknowledged as an author and its generated works are granted copyright protection, it implies that “human creativity” and “machine creativity” are regarded equally. Conversely, if AI-generated works lack copyright protection, it suggests a preference for human creativity over that of machines. Elevating machine creativity to the same level as human creativity, or favouring it, could ultimately undermine human creativity in the long term.
Recognizing AI as the author of its generated works may lead to various complications. The output produced by AI may not be without flaws. It could potentially employ biased or harmful language, resulting in defamation or obscenity, inciting violence based on caste, creed, or religion, or generating other undesirable outcomes. In such instances, attributing civil and criminal liability to the AI becomes challenging, as it is not considered a legal person. The most that can be done is to delete the problematic work or, in severe cases, ban the AI software; however, by that time, irreversible damage may have already occurred. Additionally, if an AI-generated work is found to be “substantially similar” to an existing copyrighted work, determining how to hold the AI accountable for infringement poses another significant challenge. Furthermore, if AI is classified as an author, it would lack the capacity to transfer ownership of the work due to the absence of personhood.
It is important to note that the copyright legislation in numerous countries grants moral rights to authors, although such provisions are not mandated by the TRIPs Agreement. Typically, two moral rights are recognized: (i) the right of attribution, and (ii) the right of integrity. The right of attribution allows authors to be acknowledged as the creators of their works, while the right of integrity permits authors to seek compensation for any alterations or distortions of their work that could harm their honour or reputation. The moral rights of an author are fundamental to their creations, granting them the authority to preserve, protect, and cultivate their works. These rights are inherently tied to the emotions and sentiments of the human author and are not applicable to artificial intelligence.
Another complex issue pertains to the duration of copyright for works generated by artificial intelligence. Unlike humans, AI does not experience mortality. One could contend that the copyright term should commence from the date of publication, lasting for either 50 or 60 years, contingent upon the specific laws of various countries. The debate surrounding the extension of copyright protection to AI-generated works arises from the notion that human creators are mortal and subject to fatigue, which limits their output during their lifetimes. Consequently, copyright is deemed appropriate as it rewards the efforts of human authors. In contrast, AI is not bound by mortality or fatigue and can generate an unlimited number of works. Thus, the question of granting copyright protection to AI-generated creations remains “ambiguous and contentious.” Furthermore, experts opposing copyright for AI-generated works argue that if the same model and identical inputs are utilized, the AI will consistently produce the same output. This raises concerns about the uniqueness and creativity of such works. Furthermore, it will be challenging for artificial intelligence to negotiate royalties with other parties and to uphold the rights granted to authors under copyright law. Designating AI as the author of a work is likely to present more complications than solutions.
Another perspective that arises from this discussion is the notion that AI-generated works should not have an assigned authorship and should instead be classified as part of the “public domain.” However, it is important to consider that if AI-generated works lack protection, the public could utilize these creations without authorization or compensation. This scenario could jeopardize companies that invest substantial resources in developing AI systems to produce such works. Individuals may begin to commercialize these creations in various ways without incurring costs, thereby competing with the companies that have made significant investments. Consequently, some form of protection for AI-generated works may be necessary to incentivize AI developers and companies, serving as a catalyst for continued investment in research and development related to artificial intelligence.
LEGAL STANDPOINT AND RAMIFICATIONS
The Copyright Act of 1957 in India establishes the legal framework for the protection of various creative works, encompassing literary, dramatic, musical, and artistic expressions. With the rapid advancement of technology, new questions have emerged regarding the ownership rights of artificial intelligence systems and their developers concerning the creation of copyrightable materials. According to the Indian Copyright Act of 1957, original works that are fixed in a tangible medium qualify for copyright protection. Although the Act does not specifically mention AI-generated content, section 2(d)(vi) indicates that the individual responsible for creating a computer-generated work is deemed the author. This provision has significant implications for AI-generated content, presenting both opportunities and challenges.
AI systems are increasingly becoming essential in the generation of new copyrightable works across a range of artistic fields. These systems, which utilize sophisticated algorithms and machine learning techniques, are capable of producing original content, including music, visual art, literature, and film scripts. By analysing extensive datasets and recognizing patterns within existing works, AI can independently create innovative and distinctive pieces. From algorithms that compose original music to models that generate visual art, the creative capabilities of AI are broadening the scope of traditional artistic expression. When we regard AI systems as creators of works based on the data they are provided, the originality of these generated works often raises concerns. Another significant factor to consider is the method by which this data is acquired. The datasets used to train AI systems must be collected legally and with the explicit consent and approval of the original authors. This issue has been underscored in recent legal actions taken by various authors of copyrighted works in the United States against a prominent AI platform. In these cases, it has been claimed that the AI platform improperly utilized the authors’ copyrighted works as training datasets. Therefore, it is necessary to find a balance, and a potential strategy to protect AI-generated creations from infringement claims, even when they draw inspiration from existing materials, which could involve classifying them as ‘derivative works’, especially considering that the Berne Convention currently has 181 signatories.
It is important to note that the Berne Convention for the Protection of Literary and Artistic Works, established in 1886, serves as an international treaty designed to create a cohesive framework for copyright protection among its member countries. Article 2(3) of the Convention defines ‘derivative works’ as new creative outputs that are based on or derived from one or more pre-existing works. To secure copyright for these derivative works under the Berne Convention, there must be a clear and significant deviation from the original source material. Although AI tools rely on data from existing sources, the outputs they generate are not mere reproductions; they are influenced by the distinct learning capabilities of the specific AI model employed. Consequently, the creations produced by AI should not be classified as copies. This complexity supports the argument that outputs generated by AI should be acknowledged as unique derivative works that reflect the AI’s insights, rather than simple rearrangements or reproductions of existing content. Therefore, works generated by generative AI models or software may be eligible for protection as derivative works, provided that the materials used to train these AI models or software were acquired legally and with the consent of the original creators.
Indian copyright law mandates that authorship must be strictly attributed to human creators. Under this legal framework, a work is only eligible for copyright protection if it is produced by a human author. Although the Copyright Act of 1957 recognizes the creators of computer-generated works, the legal standing of content generated by artificial intelligence remains ambiguous. Furthermore, the duration of copyright protection for original works in India extends for 60 years following the death of the author. If AI systems were to be recognized as authors, the ongoing existence of these systems would pose a challenge to the fundamental intent of this protection period, as they would persist indefinitely. Another significant challenge facing the intellectual property community is the enforcement of copyright laws. Assigning authorship rights to AI systems raises issues regarding the enforcement of copyrights and the accountability of these systems for potential infringements. Since AI systems do not possess legal personhood, they cannot be held liable for violations or any associated legal consequences, thereby complicating the situation further.
The recent increase in AI-generated content prompts complex inquiries regarding authorship, ownership, and enforcement of such works, as current legal frameworks endeavour to adapt to a landscape where machines play a significant role in the creative process. The inevitable convergence of AI and existing copyright law in India calls for a thorough examination and refinement of the pertinent legislation to tackle the challenges and opportunities that arise from AI’s emergence as a powerful creator of original, copyrightable works.
The report from the Parliamentary Standing Committee has suggested the establishment of a distinct category for the protection of AI-based inventions under intellectual property rights. The Committee has acknowledged the significance and practicality of advanced technologies such as AI and machine learning, particularly in their contribution to India’s economic growth. Furthermore, it has recommended that the Department for Promotion of Industry and Internal Trade undertake a review of existing laws, specifically the Patent Act of 1970 and the Copyright Act of 1957, to ensure that they encompass the innovations and creations associated with AI.
RELEVANT CASE LAWS
The RAGHAV Case
In 2020, the Indian Copyright Office encountered a significant case concerning an artificial intelligence system known as ‘RAGHAV,’ which sought copyright registration for a piece of artwork titled ‘Suryast.’ The initial application was denied due to the absence of a human author. However, the painting eventually received copyright protection when a human individual was designated as a co-author with ‘RAGHAV.’ Subsequently, the copyright office released a notice to retract the registration, acknowledging that it had erroneously approved it. The office requested that Mr. Sahni, the human co-author, consider the legal implications surrounding the AI system RAGHAV. Although the application is still listed as ‘registered’ on the copyright office’s website, the court has not yet decided on this matter.
Eastern Book Company v. D.B. Modak
The benchmark for originality in India was set in the case of Eastern Book Company v. D.B. Modak. In this instance, the Supreme Court Cases (SCC) journal, published by the Eastern Book Company (EBC) since 1969, included all Supreme Court rulings along with certain enhancements such as paragraph numbering, headnotes, formatting, and cross-references. In 2004, the respondents introduced software that contained reproductions of judgments taken directly from SCC. In response, Eastern Book Company initiated a copyright infringement lawsuit, claiming that the “original” components had been copied. The respondents contended that since government documents are not subject to copyright, this principle should extend to judicial decisions, thereby asserting that SCC held no copyright over them. The Supreme Court examined various definitions of “originality” to establish the standard applicable in India. It determined that the ‘Sweat of the Brow’ standard was insufficient for providing copyright protection, as it excessively favoured the rights of the owner. The court also considered the American standard of ‘modicum of creativity,’ which requires a certain degree of creativity for a work to qualify for copyright. However, it found this standard to be excessively stringent. Ultimately, the court adopted the Canadian test, which stipulates that the work must result from the author’s skill and judgment, ensuring that the exercise of such skill and judgment is not so trivial as to be merely mechanical. The court concluded that the human skill and judgment involved in creating the additional elements required legal expertise, skill, and the author’s discretion, thereby affirming that the SCC version of judgments is indeed copyrightable.
SUGGESTIONS AND CONCLUSION
Acknowledging the creative potential of artificial intelligence may necessitate revisions to intellectual property legislation. One suggested approach is to implement comprehensive user agreements that empower developers to delineate and negotiate ownership rights for AI-generated works. Intellectual property laws require updates to effectively govern creations produced exclusively by AI.
AI systems engage in a wide array of functions, yet the accountability for works generated by AI presents significant challenges. Determining liability for legal issues, such as copyright violations or breaches of privacy, has become increasingly complex. It is essential to establish definitive legislation regarding accountability for harmful actions performed by AI, particularly when it operates independently. This underscores the necessity for a clear legal framework. Legislation should guarantee that humans retain control over AI decision-making, ensuring that creators are held accountable and subject to penalties for errors made by AI.
Cases like Sophia’s citizenship in Saudi Arabia prompt critical discussions about the recognition and legal standing of AI entities. A balance must be struck between commercial interests and public welfare, taking into account intellectual property laws and liability issues.
Furthermore, ethical considerations surrounding AI-generated content, including bias and fairness, must be integrated into policy frameworks. The development of AI should prioritize positive contributions to society.
The creation of policies and standards governing AI creations requires international collaboration. There is a pressing need to establish global standards for content generated by AI.
The influence of artificial intelligence (AI) is set to expand significantly across various sectors of our daily lives. Consequently, there is a pressing need for legal frameworks to govern its applications. In the realm of intellectual property rights, particularly concerning copyright, AI will assume a crucial role. The challenges surrounding authorship and ownership of works produced by AI have compelled the global community to seek a universally acceptable resolution. Currently, there is no definitive solution to this dilemma, and each proposed rule presents its own shortcomings. Granting non-human authorship to AI-generated works could lead to substantial consequences. Additionally, placing these works in the public domain may deter AI developers and companies from investing further in this field. The World Intellectual Property Organization (WIPO) is actively engaged in addressing these challenges. Certain provisions within national copyright laws, specifically tailored for AI and its outputs, may help mitigate these issues. Ultimately, AI-generated works should receive limited protection, with a preference for human creativity over that of machines. Thus, a balanced approach is essential at this juncture.
AUTHOR NAME: Raj Sanjay Mitra
UNIVERSITY/ COLLEGE/ INSTITUTE: Law Centre-II, Faculty of Law, University of Delhi
CURRENT YEAR OF STUDY: Second (Fourth Semester)
COURSE: L.L.B. (3 Year)
CONTACT NUMBER: 8766390629
E-mail ID: rajsanjaymitr@gmail.com