Intersection of AI and Copyright Law

Author(s): Mehak Mathur

Paper Details: Volume 2, Issue 4

Citation: IJLSSS 2(4) 16

Page No: 151 – 165

ABSTRACT

The role of AI is rapidly expanding across all sectors of society, and its impact on our daily lives is set to grow exponentially. As AI continues to play an increasingly pivotal role, there is an urgent need for legal frameworks to govern its applications. In the realm of IP (Intellectual property), particularly copyright law, AI will undeniably play a central role. The issues relating to the authorship and ownership of AI-generated works are compelling the international community to seek a universally acceptable solution. However, there is no perfect rule to address this dilemma, and every potential solution comes with its own set of challenges.
Granting non-human authorship to AI-generated works could have significant consequences, and placing works generated by AI in the public platform may not be the best approach, as it could discourage further investment by AI programmers and the companies developing these technologies. The World Intellectual Property Organization (WIPO) is actively working to address these concerns, and a possible solution may lie in a sui generis system or tailored provisions within national copyright laws specifically addressing AI and AI-generated works.

Ultimately, any legal approach should provide less protection to AI-generated works while prioritizing human creativity over machine-generated output. A balanced approach is essential to navigating this complex issue effectively and ensuring that both innovation and human authorship are adequately protected.

CHAPTER 1: INTRODUCTION

Artificial intelligence is a latest emerging and most discussed technology these days. The term consists of two words artificial which means “something which is near to real” and intelligence means “the ability to learn, analyze, resolve, and create and anything that requires skills”. Artificial intelligence can be understood as a program which is capable of creating, analyzing, solving complex to complex problems and so much more within a click. AI includes Machine learning, natural language processing, and robotics, offer transformative potential across various sectors. Built on statistical algorithms that enable processes like machine learning and deep learning—one component of AI’s complex structure—there is considerable hope that this technology will boost efficiency, effectiveness, and contribute positively to societal, economic, and social welfare. AI is expected to tackle global issues, such as addressing climate change and the COVID-19 pandemic, and to enhance the functioning of public and private organizations on both small and large scales.

Artificial intelligence has changed the shape of various sectors including legal. It is a disruptive technology as it has replaced existing technology and methodology by solving advanced problems. It has resulted into lay-offs of existing employees, as most of the people have shifted their operations to AI. Artificial Intelligence (AI) revolutionizes various industries by solving complex problems of advanced problems. It is an effective and cost-efficient way in which companies can generate roaring profits. AI reduces manual labor, time, and efforts. AI in the legal industry is still in its infancy, but its adoption is rapidly increasing. According to a report by Grand View Research, the global legal AI market size was valued at $714.4 million 1077 in 2020 and is expected to grow at a Compound Annual Growth Rate (CAGR) of 37.9% from 2021 to 2028 (Grand View Research, 2021). The report attributes this growth to the increasing demand for automation in legal processes, rising legal expenses, and the need for efficient contract management.

AI has surely changed the landscape for the world but it has serious repercussions too. It can be easily misused and increase crimes in the country. Elon Musk, the most successful entrepreneurs of all the time, founder of Space X, co-founded many companies, owner of X (Formerly Twitter) , in an interview said that AI is dangerous than a nuclear bomb .Ever since AI tools have emerged there has been a serious issue of deep fakes, impersonation and frauds all over the country. With the misuse of AI, there has been an increase in the cybercrimes and e-commerce frauds. The legislation of India has taken steps in order to protect data, yet, the technology is so advanced that the tracking of the perpetrators is next to impossible.

AI has caused challenges like data privacy, ownership issues, Copyright violations, unemployment and has made people lazy. This paper will be focused on understanding the legal implications of AI on copyrights, and persisting legal frameworks and challenges of this rapidly evolving field.
This research paper focuses on analyzing the artificial intelligence, its models, benefits and challenges relating to Intellectual Property, specifically copyright laws.

LITERATURE REVIEW

There is a substantial amount of research papers, articles and blogs on Artificial Intelligence, and its impact on performing tasks, replacing human capital and its threats, specifically in the field of Intellectual Property Rights. Also relevant to this paper, is previous work establishing a relationship between Artificial Intelligence and Copyright law. The existing literature on the topics Artificial Intelligence and Copyrights have been reviewed in this paper.

STATEMENT OF PROBLEM

In the era of emerging technologies, artificial intelligence stands as a disruptive force. While various AI tools like ChatGPT, MetaAI, and others simplify our daily tasks, they also bring potential risks to society. There is a fear of layoffs, Misuse of data and information, and rise in cybercrimes with this technology.

In the initial days of AI software, there were surges in generation of deepfake images, voices, and messages, which were used to scam people. Nowadays, there are numerous cases of money scam, wherein the cyber fraudster, with the assistance of AI tools, mimic the voice of the relatives or family members of the victims, and asks for the money.

Another threat by this technology is the ownership issues. There are certain works generated by the AI which are ultimately derived from the combination of data fed to the software during its training and testing. This poses a threat to copyright and ownership issues. There is a serious need of a regulatory mechanism to deal with these crimes.
This paper discusses the concerns relating to copyright and ownership of the data generated by AI.

RESEARCH OBJECTIVES

Through this research paper, we will get a comprehensive understanding of the following-
• Artificial Intelligence(AI)
• Models of AI
• Conflict of AI and Copyright
• Analysis and suggestion about the same

RESEARCH QUESTIONS

In the research paper following questions are answered-
• What is an Artificial Intelligence (AI) ?
• What are the models of AI?
• How is AI in conflict with the existing Copyright laws?
• What are the challenges, and how can these challenges solved?

HYPOTHESIS

It is to be hypothesized that the emergence and development of artificial intelligence poses a threat to the existing copyright and cyber laws.

RESEARCH METHODOLOGY

The research paper has been crafted by employing the qualitative method of research. This approach comprehensively throws a light on the Role of Artificial Intelligence, its challenges with the copyright laws, international perspective to tackle the problem and suggestions regarding intersecting laws with AI

CHAPTER 2: ARTIFICIAL INTELLIGENCE

In 1956, the Dartmouth Conference, convened by John McCarthy and others, birthed the term “Artificial Intelligence” and set forth its ambitious goal: to make machines simulate aspects of human intelligence (McCorduck, 2004) . There is no legal statute which defines the term “Artificial Intelligence”; it can be understood as the ability of machines to perform functions for which people require knowledge and skills. Ray Kurzweil defined AI in 1990 as “the science of making computers do things that require intelligence when done by humans” . Artificial intelligence generally refers to the capability of machines to carry out cognitive functions, including reasoning, perception, learning, problem-solving, and making decisions. According to Russ Pearlman, “the central goals of AI include reasoning, knowledge, planning, learning, natural language processing (e.g., understanding and speaking languages), perception, and the ability to move and manipulate objects”. The three categories of AI systems identified by WIPO are – (i) “expert (or knowledge-base) systems”; (ii) “perception systems”; and (iii) “natural language systems”.

AI enables machines to perform tasks independently or with minimal human assistance, tasks that would typically require human intelligence. Rather than a single technology, AI is a broad field encompassing various subfields, including machine learning, robotics, natural language processing, and deep learning. At the core of AI are artificial neural networks—systems modeled after the human brain’s learning processes. These networks have self-learning capabilities, enabling them to deliver improved results as they are exposed to increasing amounts of data. AI is not to be considered as one technology, rather a field which has many subfields “such as machine learning, robotics, language processing and deep learning”. “Machine learning” and “deep learning” are therefore, two components of AI.

In machine learning, a computer program uses an inbuilt algorithm that enables it to learn from data inputs and adapt to make future decisions, either independently or under guidance. Essentially, machine learning algorithms use input data provided by the programmer to create new outputs by making autonomous decisions. Thus, while the programmer establishes the parameters, the AI produces the final work. Some of the prevalent examples are- “chess playing computers to self-driven cars”, can be seen heavily relying on “deep learning” along with “natural learning process”.

Artificial Intelligence has been interwoven in our society, as it lends support to various sectors in almost every situation, from major issues to minor issues. This ever-evolving technology has been embedded in smart phones as well, to assist the user with daily tasks like typing a message, drafting an email, selecting clothes using AI tools for a party, in cars to avoid accidents, keep a record of diverse transactions that takes place in any bank or institution, assists doctors to diagnose the diseases and perform surgeries, etc. In other words, AI has been intersected in almost every field today.

Professor Stephen Hawking once said that “the development of full artificial intelligence could spell the end of the human race”. He further said that “it would take off on its own, and re-design itself at an ever-increasing rate” and “humans, who are limited by slow biological evolution, couldn’t compete, and would be superseded”

The role of AI has been increased in the area of creativity and innovation. OpenAI, an artificial intelligence lab in the United States, introduced an AI System called ChatGpt 3, which was capable of writing poetry, generating tweets, responding to trivia questions, summarizing emails, “translates languages and even writes its own computer programs. It understands the human behaviour, responds to any situation, to tackle any kind of problem. Recently, this AI system has been advanced to ChatGPT-4.0, which is a more and advanced version of the latter system. ChatGpt 4 is so much advanced version of 3rd generation that the OpenAI called it an “early (finished yet incomplete) version of an Artificial General Intelligence(AGI) system.”

MODELS OF AI

An AI model can be understood as a program, which has been subjected to the training by feeding various set of data, to recognize certain pattern and make decision without the intervention of a human being. Artificial intelligence models use different algorithms to different level inputs, to achieve outputs or tasks, for which they have been programmed for.

With the growing complexity and versatility of AI tools, there is an increasing need for substantial data and computational power to support their training and operation. To address this, traditional single-task systems are being replaced by foundation models, which are pre-trained on vast, unlabeled datasets and can be applied across multiple domains. These adaptable models can then be further refined for specific applications.

The two terms, used in AI formation are algorithms and models. These terms are used interchangeably; however, they do not mean the same. Simply breaking down, an AI Model is used to make decisions or give output, whereas Algorithms are the mathematical or pseudocodes on which the models run. Various models are of AI are enumerated below with their brief explanations-

Discriminative models: Discriminative models, often based on supervised learning, focus on identifying the boundaries between different data classes (or “decision boundaries”) in order to predict the conditional probability P(y|x) of a data point (x) belonging to a particular class (y).
Generative Model: Generative model, are based on supervised learning, model the distribution of data points, aiming to predict the joint probability P(x,y) of a given data point appearing in a particular space.
Regressive Model: Regression models are used to predict continuous values, such as price, age, size, or time. They are mainly employed to assess the relationship between one or more independent variables (x) and a dependent variable (y), with the goal of predicting the value of y based on the given x.
Classification Models: Classification models are used to predict discrete values and are primarily employed to assign labels or categorize data. These models can be binary classification tasks, such as “yes or no” or “accept or reject,” or multi-class classification tasks, like a recommendation system that suggests options such as Product A, B, C, or D.
Apart from the above models, there are Training models, Foundation models, Testing AI models, and Deploying AI Models, that processes data, analyse it in different methods or algorithms, and gives the desired output.

CHAPTER 3: ARTIFICIAL INTELLIGENCE AND COPYRIGHT

Copyright law is a type of intellectual property right, which protects the creativity of an individual. Copyrights can be extended to authors for their literature work. The literature work has a wide ambit, includes not just books, novels, and articles but is also extended to computer software, codes, tables, collections, datasets, music, lyrics, audio, cinematograph films, etc. In other words, we can understand that the law, not just safeguards the idea themselves but the manifestation of ideas. Copyrights are the rights granted to authors, musicians, painters and other artists and their creativity.

In India, copyright is governed by the Copyrights Act, 1957. Section 13 of the copyrights act, 1957, protects the literary works, audios, computer software, all kinds of creative works. Section 14 of the act, pertains to a certain exclusive right that the owner is granted after he gets the rights of the property. Only the copyright owner, or those who have explicit permission from the copyright owner, is authorized to use the work.

The objective behind this law is primarily to compensate for the labour, ideas, and efforts of the owner of the work. In order to do this, copyright guarantees the rights of authors on their creative work, while allowing others to openly expand study and review the concepts and knowledge of the works.

Computers have been extensive used since 1970s, for generation of copyrighted works. Earlier, the computers were just used as tools to create works for the artists. Computers played the role of stationary for the artists. So, it was easier to protect and provide such original and authentic works.

With the emergence of Artificial Intelligence, this has become a major problem. The artificial intelligence, works on various models and algorithms which has been discussed in the previous chapter. To get some work copyrighted, the work must be originally written, crafted or the idea must have been originated from that very author. However, with AI, this seems unreal.

The AI can create a plethora of work, in less time and investment. The works created by AI may qualify for copyright protection in all the jurisdictions for being original. The requirement of use of “skill and judgment” in originality may be deemed to have been satisfied by virtue of the “programming and parameter on which such AI actually compiles and creates the work”.

The works generated by AI can be of two types, AI Generated and AI assisted. The difference between the two is quite clear, one is wholly generated by AI and the other includes human assistance. When the work is generated by AI, there is no author, however, in the AI assisted work, human intervention is involved. The dilemma lies in within the ownership of the works generated by AI. The only way through this dilemma is- (a) The copyright system can be recognized for authorship, (b) Copyright must not be granted to the AI generated work, and such work would fall in the hands of public, (c) There must be some other statute instead of copyright right act.

CHALLENGES WITH AI WITH RESPECT TO COPYRIGHT LAWS

The copyright, serves as a motivation, encouragement for the authors to produce more creative works, using his efforts, time, skill and judgment. The chain of confusion starts with that, if AI is given the status of author, then copyright would be granted to a work generated by a machine, and in such scenario “machine creativity” would be given the same status as “Human creativity”. On the contrary, if human creativity is preferred over machine creativity. In any of the above case, human creativity and authenticity would be jeopardized.

Considering AI has an author might cause several problems. The work produced by the AI can be biased, and unethical. It is ultimately produced by a machine, which would cause no error. AI is not aware of the social ethics so, it may produce defamatory and obscene work, which would have the potential of inciting public unrest and violence. In such cases, there would be no one responsible and liable for the civil and criminal liabilities. Another potential problem which it could pose is of the plagiarism. If any AI- Generated work is detected as “substantially similar”, then there would be no face or person responsible for it.

Another issue which the AI generated work can face is of the “legal personality”. The copyright law of all the countries, points at the author’s personality in the work. This essential ingredient cannot be satisfied with AI generated work, as it does not have any legal personality. It does not have rights like a legal personality holds. Making AI a legal entity would mean that it should possess the capacity to enter into contracts with other persons. It will also have duties under the law and will be liable for its acts. Most importantly, it should have the capacity “to sue and be sued” under the law. Most of the countries are not in a favour of granting legal status to AI.

Another confusion which arises in providing copyright to the AI Generated work is regarding the Term of AI generated work. A human being is moral, so the fatigue and saturation point of ideas and thoughts is a natural phenomenon. Human generated are given copyright, in order to provide them social and economic rewards for their hard work, and recognition in the society during their lifetime. However, AI is an immoral system, there exists no such problems as human has, so copyrighting every work would also be a problem.

Further, Human generated works, as discussed can be used with the permission of the owner, along with the fee in exchange of using their creation called “Royalty”. Humans can negotiate the prices of the same, but the same cannot by the AI system. Humans can enter into contracts regarding the same but, AI cannot as it is not a legal personality.

Another issue with AI generated works is that, since it doesn’t have a personality, and is fed only the data, it doesn’t has morals, conscious of that of a human. Humans themselves can misuse it and cause destruction to many lives. These days there is a persistent issue of deep fakes and Voice scams, which ruins the reputation of various people and will continue to be destructive as this technology cannot be held liable for these criminal and civil crimes. The perpetrators, can easily use AI as a mask, to shield them from liability of the crimes.

These are certain challenges with AI generated works. To tackle these problems and challenges, efforts are being made by all the organizations and states on the international level. In India, the progress of the same is slow, as compared to the other nations like Europe, Russia and China. If some inspiration is taken from these nations, India is capable of forming its own statute relating to AI and its potential risks. However, the technology is still under development, and it is difficult to tap and count the number of risks associated with it, as even if the laws are made on the existing issues, new issues would emerge without a blink.

CHAPTER 4: ANALYSIS OF INDIAN LAWS

The Copyright Act, 1957 and AI generated works, has a complex relation, is what we can gather from the above two sections. AI is now a deeply rooted technology, as it assists, creates, does work for various people and organizations. The major issues with this technology are data protection and ownership and copyrights law. Since it is an under developing concept, and a relatively new concept, there are no precedents that can help us in this situation.

Under the Copyright Law, copyright protection can only be granted to the original content. Copyright is granted to the “author” or creator, who has exclusive rights to reproduce, distribute, and display the work, among others. Under Indian copyright law, ChatGPT or similar AI cannot be considered an author, as copyright can only be claimed by a natural person. According to Section 17 of the Indian Copyright Act, authorship is limited to “persons,” a term generally understood to mean individuals. Although entities like companies may hold copyright by assignment (Section 18), this assignment must be from an individual, based on an agreement and limited in duration. Section 17 also stipulates that, in the absence of an agreement to the contrary, the human individual who created the work will retain the initial copyright ownership. Additionally, the Act is explicitly human-focused; for example, Form-XIV, the application for copyright registration, requires the claimant to provide personal details like name, nationality, and address.

The developers’ entitlement to copyright would primarily hinge on their Terms of Use policy. If developers state within their terms that they will retain rights to content generated by their AI, they may claim copyright, barring any prior contractual agreements to the contrary. However, the Terms of Use for popular AIs like ChatGPT and BingChat do not claim copyright over user-generated outputs. Logically, it may also seem inequitable to permit developers to hold copyright over AI-generated outputs. A comparable scenario is the use of Microsoft Paint to create a painting: while Microsoft provides the tool and certain functionalities (color filling, shapes, etc.), it would be unreasonable for Microsoft to assert copyright over the artwork produced. The creative product is fundamentally the result of the user’s application of skill and creativity. Similarly, OpenAI, by providing a service, enables users to generate content, but the final product is shaped by the user’s unique direction and choices.

To balance the principles of protecting human involvement in creative processes and acknowledging the copyrightability of AI-generated content, we propose a “Significant Human Input” test. This straightforward test examines the degree of human involvement in creating an “original” work. Importantly, this is not intended as an entirely new framework for assessing copyrightability; rather, it operates within the established “Skill and Judgment” criteria, adding a further question to assess whether the “original” work would exist in its current tangible form without human intervention (further elaborated in the following section). If a work meets the “Significant Input” standard defined by this test, it may be eligible for copyright protection. This test comprises two parts: (a) determining whether the AI-generated work is “original” and (b) evaluating whether the human involvement in its creation was substantial enough to warrant protection.

On bifurcating the test, we get two elements, a) Originality of the work, b) Significance “Input” by the creator

A) ORIGINALITY OF THE CONTENT

One of the primary arguments against granting copyright protection to AI-generated content is the alleged lack of originality. Critics argue that generative AI, being incapable of independent thought, produces outputs that merely amalgamate existing material from the internet. This leads to the assertion that AI-generated content cannot be copyrighted. However, if we examine the standard of “originality” required under Indian law for copyright protection, this argument loses strength.

The current originality standard in India was established in Eastern Book Company v. D.B. Modak. In this case, the Eastern Book Company (EBC), which has published the Supreme Court Cases (SCC) journal since 1969, included paragraph numbers, headnotes, formatting, and cross-references as additional features in their reports of Supreme Court judgments. In 2004, the respondents introduced software containing identical copies of these judgments as published in SCC. EBC filed a copyright infringement suit, claiming that these additional features constituted “original” elements. The respondents argued that, as government documents are not copyrightable, this principle should also apply to judicial decisions, and therefore SCC had no copyright over them.

The Supreme Court analyzed various standards of “originality” to determine India’s benchmark. It found the “Sweat of the Brow” doctrine too lenient, as it prioritized the effort of the creator over the actual creative content.

The court also reviewed the American “modicum of creativity” standard, which requires a certain level of creativity for copyright protection, but found it too demanding. Ultimately, the court adopted the Canadian “Skill and Judgement” test, requiring that the work be a product of the author’s skill and judgment in a way that is not merely mechanical or trivial. The court found that the additional elements in SCC’s publication involved substantial legal knowledge, skill, and judgment, thereby making the SCC version of judgments copyrightable.

B) “SIGNIFICANCE INPUT”

It is also essential to consider that the Copyright Act mandates a human author. In typical cases, AI-generated works involve minimal human involvement, with much of the creative and labor-intensive work performed by the AI itself. Most of the requirements outlined in the D.B. Modak standard—such as the application of skill, knowledge, and judgment—are fulfilled by the AI rather than the human user. This presents a challenge to meeting the legal criteria for originality, as the contribution of the human is often limited to providing simple prompts or instructions, leaving the bulk of the work to the AI.

The “Significant Input” test sets out two key criteria to assess whether an author who used AI assistance in creating a work can claim copyright. First, it requires an objective examination of whether any human involvement was present in the creation process. Second, it evaluates the extent of that human involvement, specifically looking at whether the level of human skill, judgment, and labor applied was substantial enough that the final product would be fundamentally different—or would not exist at all—without such input. This case highlights that novelty is not a requisite for copyright in India, which follows a more balanced standard of originality. Consequently, there is no reason to apply a different standard to AI-generated works. As long as AI-generated content meets the required threshold of “originality” through the human-guided application of skill and judgment, it can be considered copyrightable under Indian law, regardless of its derivation from existing bodies of knowledge

THE ZARYA STANDARD: US PERSPECTIVE

Recently, the United States recognized a similar right in the Zarya of the Dawn (‘Zarya’) case. In this case, the U.S. Copyright Office granted protection to a graphic novel, even though all the images were generated by AI. The decision was based on an evaluation of the skill, judgment, and creativity demonstrated by the author in arranging, selecting, and editing the AI-generated images. This human involvement ultimately transformed the images into a cohesive, tangible product—the graphic novel. In response to this case, the U.S. Copyright Office issued a formal policy statement, stating that a work containing AI-generated material could be granted copyright if it includes sufficient human authorship that meets the required standards for copyright protection.

Indian Judiciary is enriched with abundance of knowledge, experience and fairness. Time and time again, the Supreme Court has tackled problems and has devised the best laws, looking at the diversification of the country. Therefore, India can take an inspiration from the laws of the countries and can make a change in the legal landscape of this “AI era”.

It will however, not be out of place to mention that the European Parliament has advocated to grant the legal status of “electronic persons” to “autonomous robots” for the purposes of protection under copyright law.

CHAPTER 5: CONCLUSION

Granting non-human authorship to AI-generated works could have significant consequences, and placing AI-generated works in the public domain may not be the best approach, as it could discourage further investment by AI programmers and the companies developing these technologies. The World Intellectual Property Organization (WIPO) is actively working to address these concerns, and a possible solution may lie in a sui generis system or tailored provisions within national copyright laws specifically addressing AI and AI-generated works.