THE IMPACT OF ARTIFICIAL INTELLIGENCE ON COPYRIGHT LAW

Author: Seleena Saju, a student at Symbiosis Law School Hyderabad.

ABSTRACT 

Artificial Intelligence (AI) is restructuring the outline of copyright law based on new aspects against the classic bases of authorship, originality, and ownership. The following article takes stock of the capacity of AI, in the process of autonomous content production, that can shake existing structures of legalism, requiring statutory amendments immediately. By perusal of relevant case laws and provisions from statutes, it navigates how courts and law-making organs have approached issues worldwide.

INTRODUCTION

Here’s a new introduction to Artificial Intelligence (AI) has evolved from being a futuristic concept to a transformative force across industries, including the creative sector. AI systems now compose symphonies, draft novels, design visual art, and even generate innovative technological solutions. While these advancements represent groundbreaking progress, they also challenge the fundamental tenets of copyright law, which was historically designed for human creators.

Copyright law, as a basic legal framework, strives to protect original works of authorship, thereby fostering creativity by creating exclusive rights for their creators. A human-centred basis is assumed through the principles of authorship and originality at the heart of this framework. However, a new kind of work that does not require direct human input – AI-generated work, which independently produces content-raises pivotal questions: Can these works qualify for copyright protection? If so, then who owns rights-the programmer, the user, or perhaps nobody at all? Legislatures and courts are starting to grapple with these issues. Computer-generated works are recognized in the United Kingdom through specific provisions. Other jurisdictions, such as India and the United States, continue to rely on their traditional principles which do not make room for creations of AI. The European Union’s AI Act represents a sweeping effort to regulate AI technologies, although again, specific questions about copyright are left unanswered.

This article unpacks in detail the intertwined nature of AI and copyright law, meeting courts, lawmakers, and stakeholders in this unprecedented territory. By analysing key case laws, legislative developments, and the underlying implications of AI-generated works, it will provide a baseline as to how copyright law may shape innovation amidst creators’ rights in the future.

ANALYSIS

The challenges posed by AI in copyright law are not hypothetical but rather real and immediate. DALL-E or Deep Art, for instance, are AI-generated art pieces already on sale. The painting, titled “Edmond de Belamy” and painted with the help of artificial intelligence techniques, was auctioned in 2018 for $432,500 (3,72,79,565 in Indian Rupees), raising questions regarding its copyright ownership. With the growing popularity of AI-created music and literature, the need to fill the legal vacuum surrounding these creations has become ever more pressing. 

Copyright law has traditionally required that works be original and authored by a human. The Berne Convention for the Protection of Literary and Artistic Works, to which most countries are signatories, makes no provision regarding non-human authorship. Absent such specific provisions, courts have generally construed the term “author” to mean a natural person.

However, AI contradicts this principle. GPT-4 and Mid-Journey are examples of AI systems that operate independently and create content with minimal human intervention. The question is: Can such works be considered “original” under copyright law? The U.S. Supreme Court in Feist Publications, Inc. v. Rural Telephone Service Co. (1991) held that originality demands a minimum level of creativity. AI-generated works arguably satisfy this requirement, but the absence of human authorship makes them ineligible for protection.

If works created by artificial intelligence are considered copyrightable, then it becomes important to address the ownership of such a work. Such ownership can be bestowed upon the programmer, the user who feeds in the prompts, or even on the AI itself. For example, in Naruto v. Slater (2018), the court decided that a monkey cannot own copyright, thereby further strengthening the principle that only human beings can be considered as authors under US copyright law. Similarly, by analogy, this ruling may be taken to mean that copyright ownership cannot be claimed by AI systems, and all rights are to be vested in their human collaborators or programmers. 

Applying this line of thought, AI systems -being non-human entities-cannot own copyrights, and rights automatically go to the creators or users of the AI. Copyright law grants economic and moral rights to authors. Economic rights allow authors to benefit from their works, while moral rights guarantee recognition and protection against derogatory treatment. The problem is that it becomes difficult to attribute moral rights to AI-generated works. How can an AI claim credit or suffer reputational harm? This opens the possibility of limiting moral rights to human contributors.

AI’s reliance on large datasets for training purposes introduces concerns about copyright infringement. Machine learning models often scrape copyrighted material without explicit permission, raising questions about fair use. In Authors Guild v. Google, Inc. (2015), the court held that Google’s scanning of books for a searchable database constituted fair use. However, the decision’s applicability to AI remains uncertain, especially when AI replicates copyrighted material in its outputs.

CASE LAWS

  1. U.S. Jurisprudence

Feist Publications, Inc. v. Rural Telephone Service Co. (1991): This case established originality as a condition precedent to copyright protection, with a minimum threshold of creativity. AI-generated works, though creative, violate this standard because they are not the product of human imagination.

Authors Guild v. Google, Inc. (2015): It sanctioned the use of a large volume of data by declaring fair use applicable to making books searchable. It blurs boundaries, which the application of that precedent presents about AI-generated works, whenever the AI uses the copyrighted work in training its programs.

  1. United Kingdom

Donaldson v. Beckett (1774): This judgment is historic because it tied copyright with authorship; thereby, affirming that rights are inherently human intellect creativity. Such precedent underlies the human-centred approach to copyright in the UK. 

Nova Productions Ltd v. Mazooma Games Ltd. (2007): It established the principle that human intellectual effort is the sine qua non for authorship and that works that are generated automatically, such as those involving artificial intelligence, are unlikely to satisfy this requirement except where a human has played a significant creative role.

  1. India

R.G. Anand v. Deluxe Films (1978): Established the test to show “substantial similarity” in cases of copyright infringement, which bears on whether or not AI-generates contents, which too closely copy existing contents, infringe those contents. 

Eastern Book Company v. D.B. Modak (2008): Originality standard reconfirmed under the Copyright Act 1957; Section 13 requires human input. Works lacking such input due to generation under AI systems immediately run counter to this framework.

  1. European Union

Infopaq International A/S v. Danske Dagblades Forening (2009): The CJEU held that originality requires the intellectual creation of the author. On this basis, the Court’s ruling effectively bars from recognition as authors in EU copyright law any and all non-human entities, including AI.

LEGISLATIVE DEVELOPMENTS

Several jurisdictions are exploring legislative reforms to address AI-generated works. Following are the example of such developments:

  1. United States: The U.S. Copyright Office has published new guidelines that state works created by an AI are not eligible for copyright protection; conversely, the human-AI collaboration might be. Furthermore, the discussions aim to amend the Copyright Act regarding the definition of authorship and how it specifically deals with AI.
  2. European Union: The AI Act of the European Union introduces a regulatory framework on transparency and accountability in AI systems. However, it does not directly talk about copyright, yet the EU Copyright Directive again reminds everyone that an author should be a human. Apparently, the European Commission is to consider some supplemental guidelines in copyright matters with the use of AI.
  3. India: The Copyright Act 1957 makes no specific provision in regard to AI-generated works. Nevertheless, the Indian government has been engaging stakeholders in discussion and is considering amendments. The latest draft proposal specifies cooperation between human and AI creators, with eligibility for copyright protection where an important contribution is demonstrable from the human element.
  4. United Kingdom: Section 9(3) of the Copyright, Designs and Patents Act, 1988, acknowledges computer-generated works but assigns authorship to the “person making the arrangements.” The UK Intellectual Property Office has opened a call for evidence to find out whether this provision remains fit for purpose in the light of advancing AI technology. Some possible reforms might be to make clear what is meant by “arrangements” in light of the current capabilities of AI.

CONCLUSION

The advent of AI has fundamentally challenged the tenets of copyright law, making it necessary for a re-evaluation of authorship, ownership, and infringement principles. Courts and legislators must strike a balance between incentivizing innovation and protecting creators’ rights. While existing legal frameworks provide some guidance, the unique characteristics of AI-generated works demand tailored solutions. 

The rapid pace of Artificial Intelligence (AI) development is throwing unprecedented challenges to the traditional copyright law principles based on human authorship and originality. While the principles, deeply rooted in human authorship and originality, are proving to be less and less relevant in the context of AI’s autonomous creative capabilities, AI-generated works challenge the long-standing human intellectual input requirement for originality, authorship, ownership, and infringement issues.

Possible reforms include:

  1. Human-AI collaboration should be recognized as a basis for copyright protection.
  2. There should be a sui generis system for AI-generated works.
  3. There should be a clarification of fair use provisions on AI training datasets.
  4. Jurisdictions should amend their copyright frameworks to specifically include AI-generated works. Introducing provisions that assign copyright to the human collaborators, programmers, or users based on their contribution may give legal certainty.
  5. Design a rights hierarchy founded on the human involvement factor. For example, where AI acts altogether independently, primary rights would probably lie with the programmers. In other collaborative scenarios where human creativity and imagination are added, joint ownership could apply.
  6. Governments can work to create a new copyright class for AI-made works. Such a class would hold that AI products shall be copyrighted without undermining the rights of the human creators whose work may be implicated in the training of the AI.
  7. Jurisdictions should realize transparency and accountability in datasets in AI training. Policies will involve licensing or fair compensation of the creators’ works involved in the dataset.
  8. International cooperation is the way forward as AI crosses borders. Bodies such as the World Intellectual Property Organization (WIPO) should take the lead in creating a harmonized framework that addresses AI and copyright.

Further evolutions in AI will amplify the impact it makes on copyright law, so these challenges have to be addressed proactively by all concerned stakeholders. And embracing such reforms will enable copyright law to reconcile the stimulation of innovation with the protection of rights of human creators, so there is a level playing field that is sustainable.

FAQ

1. Can AI-generated works qualify for copyright protection?

Current works by artificial intelligence do not have a claim to copyright. Most jurisdictions require that works be created by human beings. For instance, the U.S. copyright law only accepts natural persons as qualified creators and excludes fully autonomous works generated by a computer. By contrast, computer-generated works in the United Kingdom are recognized with specific provisions. The person who “arranged” its creation is identified as the author. However, jurisdictions like India and the European Union do not say much about direct provisions for AI. So long as debate continues, most AI-generated works are excluded from traditional copyright frameworks based on the fact that they were not created by humans.

2. Who owns the copyright in AI-generated works?

Ownership of AI-generated works remains an open question due to the lack of clear legal rules. A claimant may be the programmer who developed the AI, the user who supplied input or prompts, or nobody at all if the work is not eligible for copyright. Most jurisdictions tend to attribute rights to human collaborators rather than to the AI system itself. This approach also aligns with the existing principles of law that support human creativity and contribution. So far, unless the laws clarify this issue, courts and other stakeholders continue to struggle with defining rightful ownership in AI-generated works.

3. Do AI systems infringe on copyright when trained on copyrighted material?

AI systems often appropriate copyrighted material to train on these systems without an explicit license, with questions of infringement arising. Again, this centres on whether that use is a “fair use” or indeed an unauthorized republication of copyright works. It is in light of such events as Authors Guild v. Google, Inc., where the court has permitted data use on large scales under principles of fair use, that treatment of AI system scraping of content remains unsettled. This creates a need to balance the rights of creators against innovation, for which clearer regulation on the acceptable use of copyrighted material in training AI is demanded.

4. How can copyright law adapt to AI-generated works?

Legislative reforms on copyright law should meet the challenges of AI-generated works. Such reforms may even create provisions regarding AI-generated content to define the rights of programmers, users, and collaborators involved in the production process. Regulation regarding the use of copyrighted data in training by AI is also a need.

Leave a Reply

Your email address will not be published. Required fields are marked *