AI-Generated Films and Copyright: Innovation or Infringement?

AI-Generated Films and Copyright: Innovation or Infringement?

As a disruptive technology, Artificial Intelligence (AI) has developed rapidly in the world of creative industries, which were previously viewed only as a futuristic concept. The emergence of AI-generated films, works of literature produced entirely or partially with the use of machine learning algorithms, is one of the most disputable phenomena. With minimal human involvement in their creation, these items raise a timely legal issue: Are they innovations to be celebrated, or are they a form of copyright infringement? Recently, the argument has escalated with the Studio Ghibli AI trend that encouraged users to ask ChatGPT powered systems to produce images and scenes in the iconic Ghibli style. Although they were praised as a source of aesthetic value, the important question these works raised was whether replicating the unique artistic expression of a studio without permission crossed the line into infringement. The paper discusses the legal aspects of such AI-generated works, including authors, copyright, derivative use, and international law and the debate about how copyright law, which emerged in the analogue age, is finding it difficult to govern an algorithm-controlled and digital world.

Copyright Law

Fundamentally, copyright law safeguards original works of authorship that are fixed in tangible form. Authorship, originality, and fixation are the key ingredients.

• India: The Copyright Act, 1957, in Section 2(d) defines the author as one who brings about the creation of a work. The courts always needed a human element of creativity.
• United States: The Copyright Act, 1976 acknowledges the right of authorship but, as the policy of the U.S. Copyright Office, explicitly excludes non-human authors. In Thaler v. the American courts once again confirmed that autonomously-generated works by AI that have no human authorship could not receive copyright protection (Perlmutter 2023).
• United Kingdom: The Copyright, Designs and Patents Act 1988, one of the rare laws to do so, in Section 9(3), attributes authorship in computer-generated works to the individual by whom the arrangements to produce the work are made. This has become a world-stopping provision in the AI debate.
• European Union: The copyright directive 2019 (EU) currently offers no particular approach to the issue of authorship by artificial intelligence, yet indicates that a human element of originality is required, as with Infopaq International A/S v. Danske Dagblades Forening (CJEU, 2009).

This lack of uniformity in jurisdictions poses a problem to AI-generated movies, which usually find their way into all parts of the world through streaming.

Who owns an AI film? Authorship/Ownership

Among the key issues is who the author of AI-generated films is.

• The Programmer: Some say that the authorship belongs to the developer, who created the algorithm, since they formed the basis of its creative outputs.
• The User/Producer: It is also said that the manufacturer of commands, edits the work, and controls the AI must be the one who owns it.
• No One: This is the hard human authorship position, embraced in the U.S., which posits that AI-generated works can be subject to the public domain.

In India, the Copyright Act defines authorship as applied in Section 2(d) to a human being, so the idea of purely AI-created films being protected is rather unlikely. But where there is significant human input-through editing, scripting or even directing, then these human inputs might be safeguarded.

Compound Works and Inventive Style

The use of derivatives is a second big issue. Numerous AIs generated films replicate the imagery of popular studios or directors. In recent times, the hand-drawn animation of Studio Ghibli has been duplicated using AI technologies, which raises the question of unauthorized copying.

Under copyright law:

• India, Section 14: Gives the copyright holder exclusive rights to reproduce or to adapt the work.
• Copyright Act, 17 U.S.C. 106: Secures against production of a derivative work without permission.
• EU Directive: has protection of reproduction, even stylistically or expressionally, of a substantial part.

An AI film that attempts to ape the style of a given artist so far as to reproduce their expression may fall under the derivative work classification, violating copyright. The question, however, before courts is whether a style in itself, rather than a direct copy, should be regarded as a protectable expression.

Copyright Law Infringement and Fair Use

Another aspect that AI films bring up is whether it is a copyright infringement to train AI using copyrighted data. The vast majority of generative AI models are trained using large collections of films, scripts, and images, including many that are copyrighted.

In the U.S., corporations claim that this usage is under fair use, particularly when so used with a transformative purpose. Nonetheless, there are pending legal cases (e.g., Andersen v. Stability AI) that disagree with this opinion.

India has no general fair use, but Section 52 of the Copyright Act, 1957, contains a number of more specific exceptions, including research, criticism, or review. The ability to commercially train AI models on copyrighted materials without their approval might thus amount to infringement.

There are text and data mining exemptions in the EU (Article 3, 4 of the DSM Directive), although rights holders may refuse, providing a degree of protection.

Therefore, AI-generated movies tend to fall somewhere in the grey area of both innovation and infringement.

The Regulatory Loopholes

One of the issues is that the existing copyright laws do not take AI into account. The law structures presuppose the presence of a human author, human creativity, and human responsibility. AI breaks all these assumptions.

• Lack of Accountability: In the case of an AI film infringing a copyright, who bears liability? The Artificial Intelligence system, the producer, or the programmer?
• Term of Rights: Copyright is thought to last decades even after the death of the author. What about the period in case of an author being an AI?
• Global Inconsistency: comparison between the UK and the U.S., and India: computer-generated work is understood in the UK, but it is not accepted in the U.S. or India, which leads to difficulties in international enforcement.

Lack of standardisation causes confusion to creators, studios, and audiences.

Striking the right balance between Innovation and Protection

Even with these issues, AI generated movies offer potential. They make movies more democratic, letting the unresourceful people create in-depth works. They also push the limits of creativity by combining human imagination and machine ability.

But in their uncontrolled form, they can undermine the rights of human creators. When worked by AI at will, the human effort in art is undermined.

Scholars of law propose the following reforms:

• Hybrid Authorship: Acknowledging the works as the collaborative effort of human and artificial intelligence, assigning the rights to the humans.
• Licensing Regimes: Copyrighted materials used as AI training datasets are mandatory to be licensed.
• New Rights Categories: A new category of rights in AI-generated works, which is a sui generis right like the database right in the EU.
• Transparency Obligations: This would mean that when AI is applied in film production, it must be disclosed.

Such steps would neither jeopardize the protective function of copyright nor its innovation.

Conclusion

The emergence of AI-created movies undermines the copyright law itself. On one hand, they constitute a radical innovation, but on the other, they also reveal loopholes in the legislation related to authorship, derivative works, and infringement.

The jurisdictions vary in terms of being human centered (India and the U.S.), computer-generated (the UK), and mid-line (EU). Until the harmonizing structures are created, any AI film will be at risk of being the motivation behind a lawsuit.

Ultimately, there has to be a change in copyright law. It should maintain its two-fold intentions of promoting creativity as well as protecting the rights of the creators. The issue is that legislators have to make sure that no matter how inventive and, at the same time, AI-aided the future of cinema may be, it can still be responsible in the legal dimension.

Statutes and Regulations

Government of India. (1957). The Copyright Act, 1957 (No. 14 of 1957). India Code. https://www.indiacode.nic.in/handle/123456789/1964?view_type=browse&sam_handle=123456789/1362

U.S. Congress. (1976). Copyright Act of 1976 (Title 17, United States Code). U.S. Copyright Office. https://www.copyright.gov/title17/

Government of the United Kingdom. (1988). Copyright, Designs and Patents Act 1988 (c. 48). Legislation.gov.uk. https://www.legislation.gov.uk/ukpga/1988/48/contents

European Parliament & Council of the European Union. (2019). Directive (EU) 2019/790 on copyright and related rights in the Digital Single Market (DSM Directive). EUR-Lex. https://eur-lex.europa.eu/eli/dir/2019/790/oj

Government of India. (1957). Copyright Act, 1957: Sections 14 & 52. India Code. https://www.indiacode.nic.in/handle/123456789/1964

Case Law

Thaler v. Perlmutter, No. 1:22-cv-01564 (D.D.C. Aug. 18, 2023). https://storage.courtlistener.com/recap/gov.uscourts.dcd.245177/gov.uscourts.dcd.245177.24.0.pdf

Infopaq International A/S v. Danske Dagblades Forening, Case C-5/08, ECLI:EU:C:2009:465 (CJEU 2009). https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A62008CJ0005

Andersen v. Stability AI Ltd., No. 3:23-cv-00201 (N.D. Cal. filed Jan. 13, 2023). https://www.courtlistener.com/docket/66686507/andersen-v-stability-ai-ltd/

Reports, Articles & Commentaries

Vincent, J. (2023, March 1). AI art is stealing the aesthetics of famous artists like Studio Ghibli. The Verge. https://www.theverge.com/2023/3/1/23619680/ai-art-studio-ghibli-style-trend

Perlmutter, S. (2023). Copyright and artificial intelligence [Scholarly commentary]. SSRN. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4374540

Ginsburg, J. C., & Budiardjo, L. A. (2020). Authors and machines [Columbia Public Law Research Paper No. 14-666]. SSRN. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3518625

Samuelson, P. (2020). Implications of AI for copyright. Berkeley Law Scholarship Repository. https://scholarship.law.berkeley.edu/facpubs/3121

World Intellectual Property Organization. (2019). Artificial intelligence and intellectual property policy. WIPO. https://www.wipo.int/about-ip/en/artificial_intelligence/