Author: Avi Bansal , a student of Maharaja Surajmal Institute, GGSIPU
Abstract
Generative artificial intelligence tools now produce text images music and other content that can compete with human creative works yet copyright law continues to rely on the idea of a human author. This article examines whether and when AI assisted outputs can be treated as works of authorship and who should own rights in such material. It engages with the policy position of the United States Copyright Office on human authorship and with key litigation including Getty Images v Stability AI on the use of protected works in training data. The discussion then considers how moral rights and the personal bond between author and work can be preserved when AI is used as a creative tool. Finally the article offers a normative view that copyright should remain centred on human creators and that AI systems should be treated as instruments rather than independent rights holders even when they play a major role in content production.
To the point
The core issue is whether outputs produced with the help of generative AI systems satisfy the basic requirements of copyright law namely authorship originality and fixation. Traditional doctrine assumes that an author is either a natural person or a legal person acting through human agents and it links protection to human intellectual effort. AI systems disrupt this assumption because they can produce complex images or narratives in response to short prompts and they rely on large datasets that include copyrighted works.
Policy analysis so far suggests that the law should not recognise AI itself as an author but should focus on the human involvement in the creative process. Where a human designs a model curates training data and shapes outputs through extensive editing or arrangement the resulting work can still be treated as a human authored composition. When prompts alone generate content with no meaningful human selection or modification the work lacks human authorship and should fall outside the scope of copyright protection. This position keeps the debate tightly focused on the threshold question of who is the author and avoids broad claims about whether machines can be creative in a philosophical sense.
Use of legal jargon and IPR concepts
Copyright law uses a settled vocabulary to describe the relationship between authors and works and that vocabulary remains helpful in the context of AI. The term authorship refers to the status of being recognised as the originator of a work and it is linked to economic rights such as reproduction communication to the public and adaptation. Originality captures the minimal level of creativity required for protection and is tied to the idea that copyright rewards intellectual labour rather than mere investment. Fixation denotes the embodiment of a work in a tangible medium such as a digital file which remains a straightforward requirement even with AI.
Moral rights including the right of paternity or attribution and the right of integrity protect the personal bond between author and work and they play an important role in many jurisdictions. Concepts such as joint authorship assignment licensing secondary infringement and fair dealing also remain relevant when courts assess how AI developers and users interact with third party rights holders. In disputes concerning training data questions of primary infringement secondary infringement and passing off have already appeared as seen in Getty Images v Stability AI. By applying these familiar doctrinal tools to AI related facts courts can develop the law incrementally instead of creating wholly new categories of rights.
The proof
The most detailed public analysis of AI and copyright comes from the United States Copyright Office which has issued a multi part report on artificial intelligence and copyright. Part two of this report addresses the copyrightability of outputs created using generative AI and states that such outputs may be protected only where a human author has determined sufficient expressive elements. It emphasises that mere provision of text prompts without further creative shaping will not satisfy the requirement of human authorship. The Office also confirms that existing legal doctrines are flexible enough to deal with AI and that new sui generis rights for machine generated content are unnecessary.
Registration guidance issued by the Office in 2023 instructs applicants to disclose AI generated material and to claim protection only for human authored aspects such as selection coordination and arrangement. Commentary on this guidance discusses review board decisions where registration was refused for images produced entirely by an AI system but granted for human written text and for the creative combination of visual and textual elements. These materials show that human creativity remains central to copyright policy.
On the litigation side Getty Images v Stability AI represents an early attempt to apply copyright and trade mark principles to AI training and outputs. Getty argued that Stability had copied millions of its photographs and watermarks to train its Stable Diffusion model without authorisation and that this use amounted to infringement and passing off. The High Court concluded that an AI model that does not store or reproduce complete works is not itself an infringing copy under UK copyright law although it found limited trade mark infringement because generated images sometimes included Getty marks. The judgment underscores the need to distinguish between training activity and individual outputs and it highlights the social importance of balancing incentives for creators and innovation in AI.
Case laws and landmark developments
Although formal case law on AI authorship remains limited there are several decisions and official positions that shape the debate. Review board decisions referenced in commentary on the Copyright Office report have rejected registration for works generated entirely by AI while recognising protection for human authored text and arrangements that incorporate AI outputs. These decisions confirm that human authorship is a threshold requirement and that AI assistance does not bar protection where human creativity is perceptible.
In Getty Images v Stability AI the UK High Court issued a detailed ruling on secondary copyright infringement and trade mark liability in the context of generative AI. The court held that Stable Diffusion as a model was not an infringing copy because it did not store or reproduce entire copyrighted works but it accepted that Getty marks appearing in outputs could constitute trade mark infringement for which the model provider bore responsibility. Commentary by intellectual property practitioners notes that the judgment provides guidance on how traditional copyright concepts apply to AI systems and may influence future disputes over training data and model design.
Together these materials form a growing body of authority that treats AI as a powerful technology whose use can give rise to infringement but does not grant it the status of author in its own right.
Human centred analysis and moral rights
Keeping copyright tied to human authorship also protects the rationale behind moral rights. The right of attribution allows an author to insist on being named in connection with their work and the right of integrity permits them to object to distortions that harm their honour or reputation. If AI systems were treated as authors moral rights would lose their personal foundation because machines have no personality reputation or dignity.
Policy analysis therefore suggests that moral rights should belong only to human creators even in collaborative contexts where AI plays a significant role. When several human contributors design prompts edit outputs and assemble them into a final work the law can recognise joint authorship and allocate both economic and moral rights accordingly. AI outputs in such cases are raw material and the real object of protection is the human effort that shapes them into an expressive work. This perspective keeps copyright close to its cultural and constitutional purposes.
Conclusion
The intersection of generative AI and copyright brings new factual patterns but it does not require abandonment of long standing legal principles. Authors must still be human and originality must still reflect human intellectual effort even if that effort is expressed through digital tools and complex models. Official guidance from the Copyright Office and early case law such as Getty Images v Stability AI show that courts and policy makers can adapt existing doctrines to AI by focusing on the degree of human control over expressive elements and by carefully examining how training data is used.
For jurisdictions including India the most coherent path is to maintain a human centred concept of authorship while dealing with concerns about training and market disruption through transparency and licensing frameworks rather than by granting rights to machines. AI should remain an instrument of human creativity and not a substitute for the human author because copyright ultimately exists to recognise and reward human intellect and imagination.
FAQ
Q1What is the basic position on authorship of AI generated works?
The prevailing position is that copyright protects only human authored works and that fully AI generated outputs without meaningful human input lack the required human authorship. Human contributions such as editing arranging or integrating AI outputs into a larger composition can still qualify for protection.
Q2Does using generative AI to assist in writing or illustration remove copyright protection?
No. Policy guidance confirms that using AI as a tool does not bar copyrightability so long as the human author determines the expressive elements of the work and their contribution meets the originality threshold. The law will focus on the human authored portions and on the way they are combined rather than on the raw machine generated material.
Q3How did Getty Images v Stability AI affect the legal view on AI models?
The UK High Court held that the Stable Diffusion model was not an infringing copy because it did not store or reproduce entire works but it did find trade mark infringement when outputs contained Getty marks. The case signalled that courts will scrutinise training practices and output behaviour while still applying conventional copyright and trade mark doctrines.
Q4Can AI ever be recognised as an author under current copyright frameworks?
Current analysis and guidance reject the idea of AI as an author because copyright is designed to protect human creativity and to grant rights to persons who can hold and exercise those rights. Any change in this position would require major legislative reform and could undermine the constitutional and policy goals of copyright law.
