The Legal Implications of AI-Generated Content Ownership


Author: Vrinda, National Law School of India University, Bangalore

Abstract

The advent of generative artificial intelligence (AI) has transformed the creation of content by producing text, images and more with an unprecedented level of autonomy. Nonetheless, this new technological advancement presents a new kind of challenge for the legal landscape, particularly the intellectual property law. The use of AI has raised challenges for the ownership and copyrightability of the AI-generated content. This article will explore whether the output generated by the Artificial Intelligence is qualified for copyright protection and who should own this work, and also the liability of the implications of the potential infringements. Through analysing the global legal framework, recent case laws and the emerging policy proposal, it focuses on the need for adaptive legislation to address this issue. The article concludes with a recommendation for the stakeholders.

To the Point

Generative AI models such as ChatGPT, Claude and DALL-E created output that can compete with human creativity from the novels to the visual arts. Compared to the human creators, AI lacks any legal personality and raises significant questions about the authorship, ownership and liability under the current IP Law. The core issues are given below:

Copyright eligibility: In conventional copyright law, human authorship is a necessity, but AI-generated works challenge this notion of IP law. The important question is whether the output of a non-human entity can be treated as an original work of authorship.
Ownership allocation: Even if the copyright applies to the AI-generated content, another challenge is who will own the work– the developer of the Generative AI model or the user who gives the prompt or idea for the generation of the content, or neither of them. Another concern will be deciding whether these works should be considered as a part of the public domain.
Liability for the infringement: If these AI-generated content infringes on any existing IP rights, who will be responsible – the developer, user or the AI (it is not even a viable option)?


Use of Legal Jargon

The legal discourse regarding the AI-generated content includes some specialised terminology. The copyright law revolves around the notion of originality, which can be understood better as a work that reflects the intellectual creation of the author. With regard to AI, this idea of authorship is contentious since the intent behind these creations rests with the human, not the machines. The concept of the derivative works in which the creation is based on some existing material applies only when AI generate some content from the copyrighted material. Using the copyrighted dataset can lead to infringement claims. The concept of vicarious liability could hold the developers accountable for the AI content which infringes the intellectual property rights, while the indemnity clauses in the licensing agreement may shift the responsibility. A sui generis right, which is different from the traditional copyright, could offer a solution similar to protection for databases in the European Union (EU). Furthermore, moral rights (like the right to attribution and integrity) raise some important questions about whether the AI users can claim such kind of protections.

The Proof

Indian legal framework

The Indian Copyright Act of 1957, under section 2(d)(vi), defines the author of the computer-generated works as “the person who causes the work to be created”. This implies that there is a need for human involvement. However, the autonomous output of the AI challenges this interpretation. Section 13 of the Act protects the “original” literary, dramatic, musical, and artistic works. The Supreme Court in Eastern Book Company v. D.B. Modak (2008) defines originality as requiring the skill and judgment that goes beyond mere labour. They also rejected the  “sweat of the brow” doctrine. This test is known as the “modicum of creativity” test. It was originated from the judgemnt of U.S. in Feist Publications, Inc. v. Rural Telephone Service Co (1991). These decisions indicate that AI-generated content needs some amount of human input to qualify for copyright protection. 

In 2021, the Indian Copyright Office registered an AI-generated artwork tool called “RAGHAV” temporarily while naming the AI as a co-author. However, this decision was later revoked because AI cannot be considered an author due to its lack of legal personality. The Delhi High Court in CBSE v. Various Publishers (2018) dismissed any copyright claims from a non-natural person. The court held that the authorship requires a human. Although no Indian court has ruled on the ownership of the AI-generated content, the decision in the case of University of Cambridge v BD Bhandari (2011) held that transformative use can limit copyrights. This decision suggests that AI training could be a fair deal if it serves a different purpose.

The ongoing case of ANI Media v OpenAI (Delhi High Court, 2024–2025) is a landmark case. In this case, the ANI alleges that OpenAI used ANI’s copyrighted news content to train ChatGPT without obtaining permission. ANI is seeking an injunction, arguing that this constitutes infringement under Section 51. OpenAI contends that its use is “transformative” under fair dealing (Section 52). This concept is similar to the U.S. concept of fair use. However, the narrower provision of India requires fair dealing for specific purposes such as research or criticism. The court has appointed amicus curiae to analyse the relationship between AI and copyright on the following issues:
1) Does storing ANI’s data for training infringe copyright?
2) Does generating responses using ANI’s data infringe copyright?
3) Is OpenAI’s use fair dealing?
The recent hearing for the case is scheduled on June 20, 2025, after an adjournment. This case also raises jurisdictional questions because the OpenAI servers are U.S.-based, and it also explores whether training AI on copyrighted material constitutes infringement. A panel from the Commerce Ministry is reviewing the adequacy of the Copyright Act for addressing AI-related disputes.
The different approaches followed by different jurisdictions are given below:
United States: The U.S. Copyright Office has denied copyright for AI-generated content that lacks human authorship. The content that involves significant human editing may qualify for copyright protection as per the 2023 guidance.

European Union: The Copyright Directive does not address the AI generative content. A 2020 report from the European Parliament suggests a sui generis system. The AI Act of 2024 mandates transparency with regard to training data, which aims to reduce infringement risks.  The 2019 Copyright Directive does not address AI-generated works.

United Kingdom: The Copyright, Designs and Patents Act 1988 (Section 9(3)) assigns authorship of computer-generated works to the arranger. However, this is an old provision that predates modern AI developments.

China: A ruling of 2020 in Shenzhen Tencent v. Shanghai Yingxun granted copyright to an AI-generated article. The IP guidelines of China in 2024 encourage the protection of AI outputs under existing laws.

Industry Practices: OpenAI’s 2025 terms assign the ownership of the generated content to the users, but they disclaim any liability for the infringement.

India’s Policy Response: The 161st Parliamentary Standing Committee Report (2021) recommended creating a separate intellectual property category for AI innovations. The 2025 Commerce Ministry panel is tasked with assessing AI’s impact on copyright and also to propose amendments. Indian currently lacks provisions similar to the EU that allow the creators to opt out of having their data used for training. It leaves them vulnerable as seen in the case of ANI v. OpenAI.

Case Laws:

Eastern Book Company v. D.B. Modak (2008): The Supreme Court established that originality requires “skill and judgment,” rather than just mere labour. This “modicum of creativity” test suggests that AI-generated works must involve significant human input to be eligible for copyright protection, in line with global standards that prioritise human creativity.

University of Cambridge v. BD Bhandari (2011): The Delhi High Court acknowledged transformative use as a limitation to copyright. This indicates that using AI for non-expressive purposes, such as pattern learning, might qualify as fair dealing under Section 52(1)(a) for research or private use. This interpretation could provide a potential defence for AI developers, although it remains untested for generative AI.

Naruto v Slater (2018, US): This case is related to a monkey that took a selfie using a photographer’s camera. The court in this case held that non-human entities cannot hold any copyright. This judgment reinforced the idea that copyright law can be applied only to human authors. This decision can be used in cases related to AI-generated content now. It highlights the strict stance of the United States on human authorship. This poses barriers to the copyright protection of the AI content.

Thaler v Perlmutter (2023, US): In this case, Dr. Stephen Thaler attempted to register an artwork generated by AI with the US Copyright Office. He tries to list AI as the author for the same artwork. The Copyright Office and the subsequent court decision denied such registration and again held that the copyright necessitates human creativity. The court further mentioned that the autonomous output lacks the “spark of human intellect”, which is required under the Copyright Act. This case presents the position of the US and also leaves open a question about the work generated by the human input.

Shenzhen Tencent v Shanghaoi Yingxun (2020, China): In this landmark decision, a Chinese court granted copyright to an article generated by the AI by attributing authorship to the human operator who gave the input to the AI. The reasoning given by the court was that the human input is sufficient to constitute a creative effort. This ruling is different from the approach taken by the West and also indicates the willingness of China to adapt the IP law to AI technology.

Conclusion

The increase of AI-generated content shows the critical gaps in the IP frameworks globally and also in India. The Copyright Act of India is rooted in human authorship, and it struggles with the idea of AI’s autonomy. The U.S. rejects the idea of AI authorship while China adapt the existing laws, and the UK legislation is outdated. The industry terms of service provide the stopgaps, but they also lack the statutory value. To address these issues, Indian and global policymakers should:

Adopt a Sui Generis rule: A framework like the one adopted by the EU in their database rights could define ownership based on the human input or the investment. This will help in fostering innovation while protecting the creators. India’s 2025 Commerce Ministry panel can pioneer this for the global South.

Redefine Authorship: The amendment in the Copyright Act to recognise the collaboration of humans and AI, which will allow users or developers to claim authorship based on their contributions. This was also suggested in the Eastern Book Company case.

Clarify the liability: There is a need to establish rules for the vicarious and the contributory violation to protect the users and the developers. The ANI case may set the precedent for training on data liability.

Enhance the fair dealing: Section 52 of the Copyright Act can be expanded to include the transformative AI training. This will help in aligning with the fair use principle of the US while preserving creator rights.

Global Harmonisation: Indian should lead the WIPO discussion to standardise the AI-IP rules. It will help in reducing the cross-border disputes and supporting India’s IT hub.

Support creators: An opt-out mechanism can be introduced, just like the EU, to allow the creators to exclude their works from AI training. It will address the concern raised in ANI v OpenAI.

The stakeholders involved in this issue must act decisively, and developers should disclose the training data while strengthening the indemnity clause. The users should also scrutinise the terms of service and avoid the infringing promotions. Indian courts and policymakers, through cases like ANI and the 2025 panel, have a unique opportunity to make a balanced framework. If the reforms are not made timely manner, it will affect India’s creative and tech sector. A proactive approach will position Indian as a global leader and also ensure equitable protection for the creators, innovators and the consumers in this AI world.

FAQS

Can AI-generated content be copyrighted in India?
No, AI cannot be an author under the Copyright Act. As given in the Eastern book company case, copyright requires human input and also the “skill and judgment test.”

Who owns AI-generated works in India?
Ownership of the AI-generated content depends on the terms of service (e.g. OpenAI assign rights to users) or the human input. Without any AI-specific laws, the dispute will persist as highlighted in the ANI Media case.

What happens if AI infringes copyright in India?
If AI infringes copyright, then the liability may fall on the users and developers. ANI v OpenAI tests whether training on copyrighted data is a fair dealing.

Are there any AI-specific copyright laws in India?
No, currently there are no such laws in India, but the 2025 Commerce Ministry panel is reviewing the Copyright Act. The 2021 Parliamentary Report also recommended a separate IP category for AI.

Can AI claim moral rights in India?
No, the moral rights require human authorship. Only the human contributors with creative input qualify under Section 57 of the Copyright Act.

What is the role of India’s 2025 Commerce Ministry panel?
The panel is evaluating the Copyright Act’s adequacy for AI. They are focusing on the ownership, fair dealing and liability. Their recommendations could lead to future amendment or a sui generis regime.

Leave a Reply

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