I. Introduction
Intellectual Property Rights (IPR) form the legal backbone for protecting creativity while ensuring that inventors, artists and writers retain control over their works which they have originally produced. But since the emergence of Generative Artificial Intelligence (Gen-AI), it challenges the very definition of originality and authorship that acknowledges these protections. Generative Artificial Intelligence encompasses sophisticated algorithms capable of autonomously producing various forms of creative output, including language, visual art and music with minimal or next to no human intervention. Generative Artificial Intelligence (GAI) refers to highly intelligent and autonomous AI systems capable of independently generating texts, images, music and other forms of creativity content tools like ChatGPT, DALL·E and Stable Diffusion operates by trainings that often include copyrighted content. This has triggered lawsuits and at the same time sparked global debates and exposed gaps in legal frameworks across jurisdictions, particularly in India.
The World Intellectual Property Organization (WIPO) defines generative AI(Gen-AI) as systems that are capable of generating outputs without human inputs. As these models create images mimicking artists' styles or text emulating published authors, they raise questions about whether such outputs are eligible for copyright and who—if anyone—owns them. Are these creations original enough to deserve protection? And if yes, should it be granted to the AI, its developer, its user, or no one at all?
II. The Copyright Challenge in India
India's copyright regime is governed by the Copyright Act of 1957 and is designed with human authorship in mind. Section 13 of the Act grants protection to "original literary, dramatic, musical and artistic works," but lacks clarity on what constitutes "originality" and whether non-human creators can qualify. It gives exclusive rights to creators; that is, the right to reproduce, distribute, publicly perform and adapt their works. While the 2012 Amendment addressed digital rights management, it did not anticipate autonomous machines producing content with minimal human input. The existing laws in contemporary times do not precisely deal with ownership and liability related issues when it comes to AI-generated content leading to legal ambiguities. In addition, it is difficult to enforce deterrent mechanisms against online copyright violation
A pivotal Indian case highlighting this ambiguity involved the painting "Suryat," co-authored by RAGHAV, an AI, and Ankit Sahni; Initially denied copyright by the Copyright Office in India the application was later amended to exclude the AI from authorship. Officials cited Section 2(d) of the Act which recognizes authorship only in relation to human intervention. But in contrast Canada recognized both RAGHAV and Sahni as co-authors. This inconsistency demands India's urgent need for AI-inclusive reforms.
III. Understanding How Generative AI Works
Generative AI models use neural networks like Generative Adversarial Networks (GANs), Variational Auto-Encoders (VAEs), or Transformers to learn from enormous corpora of data and generate new content. The models identify patterns and features in the data—often scraped from publicly available internet sources—and use prompts to create outputs that appear novel but are heavily reliant on the input datasets.
Because these datasets contain protected works, the process is fraught with allegations of copyright violation. Critics argue that if AI learns by training on copyrighted material without consent, it effectively constitutes mass-scale infringement, even if the outputs themselves are not verbatim copies.
IV. The Originality Debate: Two Legal Doctrines
Courts often use two doctrines to assess the originality:
- Sweat of the Brow: Recognizes the skill, labour, and effort involved in creating a work, regardless of creativity. This doctrine is largely obsolete.
- Modicum of Creativity: Requires that the work exhibit minimal creative input or intellectual effort.
AI fails both. It lacks subjective creativity and emotional depth; it neither intends to create nor understands what it creates. AI output may seem new but is functionally derivative, raising the question: can a machine-made collage of existing works ever be truly "original"?
V. Copyright Infringement and Liability
The attribution of liability in AI-related copyright infringement is legally ambiguous; raising complex questions about the roles and responsibilities of developers, end-users and data providers. Resolving these challenges is critical to ensuring both fair use and the integrity of intellectual property protections. AI-generated outputs face a huge backlash from many creators due to their violation of IPR of human creators, using copyrighted material in the training and output creation of AI without permission from the original authors. This has led to legal action, with companies like GitHub, Microsoft, and OpenAI facing a class-action lawsuit accusing them of engaging in 'software piracy on an unprecedented scale' through GitHub Copilot's AI-generated coding. Generative AI's capabilities have already landed tech giants in legal trouble: Following are the few examples.
- GitHub Copilot (Microsoft & OpenAI) is accused of violating software copyrights by reproducing code without attribution.
- Getty Images v. Stability AI alleges billions of unauthorized image copies.
- Drake & The Weeknd's AI-generated song went viral, prompting Universal Music Group to demand takedowns, calling it a violation of contractual and copyright rights.
In India, such disputes remain unaddressed due to a lack of tailored laws. The Copyright Act doesn't specify liability for non-human entities. Current law only holds human actors accountable—leaving questions about whether the developer, user, or the AI platform bears legal responsibility.
VI. Global Approaches: Lessons for India
- United States
The U.S. Copyright Office (USCO) explicitly denies protection for AI-generated works lacking human input. However, it recognizes AI-assisted works, where a human significantly modifies AI output. For example, in Zarya of the Dawn, a graphic novel by Kris Kashtanova, the USCO withdrew protection for AI-generated images but retained it for text arrangement and storyline, acknowledging human contribution.
- United Kingdom
Under Section 9(3) of the Copyright, Designs and Patents Act, 1988, the UK grants copyright to the person who "makes the necessary arrangements" for the creation of computer-generated works. This typically benefits developers or users. Though it doesn't recognize AI as an author, it at least allocates authorship for practical protection.
- European Union
The EU's approach is evolving. The AI Act (2024/1689) mandates that datasets used to train Gen-AI must obtain licenses when using copyrighted material, and AI providers must publish summaries of their training data. This balances innovation with artist protection and ensures AI models comply with copyright law irrespective of training location.
- China
Courts like the Beijing Internet Court have ruled that if an AI's output reflects meaningful human investment, copyright may be granted to the human user, treating the AI as a tool rather than an author.
VII. The Ethical Landscape: Creator Backlash vs. Democratization of Creativity
Proponents of generative AI argue that it democratizes creativity. Anyone can now create art, music, or essays within seconds. But for human creators, this is a threat—not just to their jobs but to their intellectual legacy. Artists report AI replicating their style without permission. Social campaigns like "NO TO AI GENERATED IMAGES" and hashtags like #SupportHumanArtists have gained traction. Musicians fear being deepfaked. Writers discover plagiarized versions of their work. These concerns aren't just theoretical. AI models have generated content featuring fragments of original watermarks, reused lyrics and melodies, or rephrased published texts—raising ethical questions about originality, theft, and artistic respect.
VIII. Why Treating AI Like a Human Creator May Be a Mistake
Some argue AI models learn like human students—absorbing data and generating novel ideas. But the analogy fails legally. Copyright doesn't protect ideas, only their expression. Humans interpret, critique, and intentionally create. AI replicates patterns. Courts have consistently resisted extending personhood to machines. Without intent or consciousness, machines cannot "create" in the legal sense.
Moreover, granting AI authorship would allow corporations to mass-produce copyrighted content with minimal human oversight, potentially devaluing genuine artistic labour.
IX. India at a Crossroads: Legal Gaps and Opportunities
Despite being a signatory to the Berne Convention and WIPO Copyright Treaty, India has not updated its copyright law to reflect the AI age. Key issues include:
- No legal recognition of AI-generated authorship
- Lack of provisions for liability
- No clarity on what constitutes "originality" in AI works
- Absence of fair use guidelines for AI training
The Indian government must acknowledge these shortcomings and begin drafting AI-specific legal provisions.
X. Policy Recommendations and the Way Forward
- Define AI-Generated Works: Amend the Copyright Act to recognize and categorize AI-generated outputs. India may adopt the UK model: assign authorship to the party arranging the creation, or define human thresholds for AI-assisted work.
- Introduce Conditional Fair Dealing for AI Training: Like the European Union's model AI training should only be allowed if the dataset is legally acquired. Additional provisions must be added that balance innovation with consent-based data use especially for regional contents and languages.
- Creation of a Licensing Model for Publishers and AI Developers: A centralized copyright society can manage licenses for AI training. This ensures compensation for creators while giving developers access to high-quality data.
- Liability Framework: Identification as to where responsibility lies developer, dataset provider or user. Consider joint liability models or obligations for watermarking AI generated content to prevent misuse.
- Public Transparency: Mandate of disclosure of training datasets and generative model architecture to assess originality and infringement risks.
- Promote International Harmonization: Align with WIPO, EU and US policy developments to ensure smooth cross-border enforcement of copyrights and data rights.
XI. Conclusion
Generative AI represents both a technological marvel and a legal issues simultaneously. While it unlocks new creative frontiers; it also simultaneously endangers the very idea of human authorship and challenges copyright law. India's Copyright Act, 1957, now faces a future it wasn't drafted to handle. As lawsuits emerges, global IP regimes evolve and human creators raise their voices; India must strike a delicate balance fostering AI innovation without sacrificing the integrity and rights of its creative community at large. One crucial aspect often overlooked in these discussions is that copyright is not an inherent natural right. It is an artificial right granted by law to promote innovation and access.
Copyright exists not merely to protect authors and publishers but also to ensure that knowledge is disseminated. Copyright law incorporates several limitations; such as duration limits, the fair dealing exception and the first-sale doctrine to maintain a delicate equilibrium between exclusive rights and public benefit. With the right legal reforms, licensing systems and ethical safeguards India can become a global leader in regulating generative AI in a way that is inclusive, forward-looking and fair.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
