The rise of artificial intelligence (AI) in contemporary times has been transformative, permeating diverse sectors such as healthcare, transportation, aviation, space exploration, education, and the entertainment industry (music, art, games, and films). AI’s automation capabilities reduce human intervention, ensuring efficiency and minimizing errors. AI’s capacity to generate local news articles, artwork, music, and even short novels has redefined creativity, allowing it to perform tasks traditionally dominated by human ingenuity, such as gaming.
However, AI’s increasing role in producing creative works, it raises complex issues in the field of copyright law. Historically, computer-generated works did not present significant copyright ownership challenges, as human input was still necessary for the creative process. These programs were considered tools—akin to stationery—used by humans to create works. However, today’s AI can generate creative works autonomously, without the need for human intervention, posing new dilemmas around authorship and copyright ownership.
The central question arises: who is the author when AI independently creates a work? In AI-assisted works, human involvement allows for clear authorship attribution. But when AI creates without human input, this becomes less straightforward. Legal systems worldwide are grappling with three potential approaches to this issue: (1) recognizing AI as the author, (2) assigning AI-generated works to the public domain, or (3) developing a sui generis legal framework to protect such works.
As AI continues to evolve rapidly, exemplified by systems like Google’s AI, which has developed and trained a “child AI,” these concerns are likely to grow. AI advancements, such as OpenAI’s GPT-3, which writes poetry, generates social media posts, and even programs, demonstrate the vast capabilities of AI in the realm of creativity. The ongoing challenge for legal frameworks globally is to balance the protection of human creativity with the acknowledgement of AI’s contributions while addressing the broader implications of AI autonomy and liability.
Who Is Considered an Author?
According to Indian copyright law, only natural persons—meaning human beings—can claim authorship of a work. Section 17 of the Indian Copyright Act specifies that authors must be individuals. While companies can own copyrights through agreements (as outlined in Section 18), they cannot be the original authors. The act is designed to prioritize human authorship, which is evident in copyright registration processes that require the claimant’s name and details. Thus, when it comes to works generated with the help of AI like ChatGPT, the question remains: who qualifies for copyright?
The developers of AI tools may lay claim to the works produced by their programs based on their “Terms of Use.” If these terms state that the developer retains rights to the content generated, they may have a valid claim. However, many widely used AIs, including ChatGPT, do not assert copyright over the outputs they create.
Prompt Giver: Can They Claim Copyright?
This is where the situation becomes more nuanced. In 1994, the Indian Copyright Act was amended to address works produced by computers, stating that the person who caused the creation of such works holds authorship. While this seems to suggest that the person providing the prompt could claim copyright, simply entering a prompt is usually insufficient for copyright protection
To qualify for copyright, a work must demonstrate originality. This means it needs to reflect some degree of human effort and creativity. For example, in the landmark case University of London Press Ltd v. Tutorial Press Ltd ([1916] 2 CH 601), the court recognized that even if a work draws from existing knowledge, the effort put into compiling it could warrant copyright protection.
However, just submitting a one-line prompt does not involve enough effort or creativity for copyright protection under Indian law. The threshold for originality must involve more than a basic input. It is crucial to examine the depth of involvement that the prompt-giver had in shaping the final output.
The Need for Extensive Datasets: Copyright Concerns
AI models are typically reliant on extensive datasets to function effectively, which often means gathering large amounts of copyrighted content from the internet. The potential for copyright infringement arises when an AI reproduces and incorporates someone else’s work into its training data. If an AI-generated output appears to infringe on existing copyrights, it likely does. While traditional copyright law holds that using someone’s copyrighted material without permission constitutes infringement, a fair use defence may be available in certain cases.
For instance, in Perfect 10 v. Google [416 F. Supp. 2d 828 (C.D. Cal. 2006)], Google was found to be infringing on copyright by creating thumbnail images from full-sized photographs. However, the court ruled that Google’s use was permissible since the thumbnails served a different function than the original images. Similarly, in the Authors Guild v. Google [804 F. 3d 202 (2nd Cir.)], the court allowed Google to create a vast database of books for searching purposes, concluding that its actions fell under fair use despite the initial infringement.
The comparison between AI behaviour and Google’s search engine operations raises intriguing legal questions. Proponents of AI argue that the processes employed by AI systems mirror the ways humans interact with data and create new works. Yet, the legal landscape is still evolving, and it remains uncertain whether specific protections or exceptions will apply to AI-generated outputs, especially considering that algorithms may operate differently than human cognition.
Legal Implications of Copyright Infringement
Even if an individual does not hold a copyright on a particular work, copyright infringement can still occur. Copyright infringement is typically viewed as a strict liability offence, meaning that the person or entity responsible for the infringement can be held accountable, regardless of intent. This liability extends to AI developers, who may face legal challenges stemming from the use of copyrighted materials in their training datasets.
Stability AI, known for its generative tools like stable diffusion, is currently facing lawsuits for allegedly downloading millions of images without proper authorization, potentially breaching UK law’s Standard of Fair Dealing. Getty Images has indicated its intention to sue Stability AI for this infringement, arguing that it seeks to establish licensing agreements rather than shut down the AI’s operations. Additionally, visual artists have filed a class-action lawsuit against Stability AI, Midjourney, and DeviantArt, claiming that these entities have violated copyright by training their image generators on copyrighted photographs without consent.
Conclusion
As the debate over AI and copyright continues, it is essential to acknowledge that the law must adapt to the realities of modern creativity. While the Stability AI case and others like it raise valid concerns about copyright infringement, they also highlight the need for clear legal frameworks that account for both the innovative potential of AI and the rights of original creators. Establishing a balance between protecting artists’ rights and allowing for the creative use of data in AI-generated works is crucial for fostering an environment of innovation while respecting intellectual property rights. The ongoing legal battles will undoubtedly shape the future of copyright law in the age of AI, guiding how we understand authorship, creativity, and the permissible boundaries of data usage.
Disclaimer: This article is for informational purposes only and does not constitute legal advice. The content may not reflect the most current legal developments and is not guaranteed to be accurate, complete, or up-to-date. Readers should consult a qualified legal professional before taking any action based on the information provided. The authors and publishers disclaim any liability for any loss or damage incurred as a result of reliance on this article. This article does not create an attorney-client relationship.