Introduction

Artificial Intelligence shot to popularity with the birth of Chatgpt. It is still in its early stages but is capable of a lot. From producing hyper realistic images to writing horror stories, AI is here to rule. AI tools like Midjourney, DALL-E and others are no stranger to us. AI voiceovers while scrolling reels, AI chatbots as customer care like Rufus at Amazon, Alexa playing bhajans and Chatgpt doing your homework. All of it is a part of our lives. Denying AI’s existence is not the way and neither is complete shut down of any use. However, appropriate regulation to ensure ethical usage of AI is the way to go. This article deals with the intersection of copyright and AI in detail.

Understanding AI Generated Art and Identifying the First Owner of its Copyright

Artificial Intelligence shot to popularity with the birth of Chatgpt. It is still in its early stages but is capable of a lot. From producing hyper realistic images to writing horror stories, AI is here to rule. The celebrity, Chatgpt is a generative AI model. Generative AI is the kind of artificial intelligence which is capable of producing ‘new’ content like texts, images, music, videos and so on. All the chatbots which help you reach your deadline by generating research papers, projects in the span of minutes are generative AIs. However, is this content really new is the matter of debate. Generative AIs like Chatgpt are able to produce new content because they are fed the data available across the internet. For instance, if Jukebox is asked to make a song like Taylor Swift it will go through her various works and then mimic her voice and make a song which will sound as if she is the one who sang it. With the advent of new developments, it has become difficult to distinguish between what is real and what is AI generated or fake.

This is one of the many ways in which AI is used to generate art. Another way which is seemingly ethical is when humans or prompt makers command AI to make art as per their instruction. In such a situation where the work is not an outright copy of someone else’s work, who should be the owner of the copyright. Section 17 of the Indian Copyrights Act, 1957 states that in ordinary circumstances the first ownership of the copyright rests with the author. Similarly, Section 2(d)(vi) states that in case of works which are computer generated, the person who causes the work to be created shall be the author. This would typically mean that the human who took the creative initiative to generate the work should be the first owner of the copyright. Section 2(d) also recognizes that an author can only be a human.

The developer of the generative AI, like OpenAI for Chatgpt, cannot be the first owner of the copyright because of the above sighted reason. The developer can still do the same if it is mentioned in the Terms & Conditions of the AI that any work developed by the AI would be the copyrighted work of the AI developer. That however would be tyrannical because the usage of a chatbot depends on the prompt giver and not the developer. When the prompt giver takes the necessary creative steps to result in the art, it is he/she who is the author and not the AI developer. The distinction between developer and employer and the non-human entity, must be noted while deciding on the matter. In the UK, the statute is similar to that of India in the way that it states that for computer generated programs, the author shall be the person who makes the necessary arrangements to cause the work. In the US, on the other hand, fully AI generated work is not recognized for copyright ownership.

Standards for Copyright Ownership

Granting copyright ownership to a fully AI generated art would seem unethical. It is still a matter of heavy debate whether AI generated art is still art[1]. However, due to paucity of space I would not be discussing on this. There are mainly two legal doctrines that are used to determine the originality of a piece of art to ultimately decide whether the art in question, is worthy of copyright ownership. The first is the ‘sweat of the brow’ doctrine. This doctrine was laid down in the case of University of London Press vs. University Tutorial Press[2]. This doctrine is used in the UK and mainly states that the labour and time put in the making of the work is more important than the level of creativity. The second doctrine is the ‘modicum of creativity’, laid out in the case of Fiest Publication Inc vs. Rural Telephone Service[3]. This is more popular in the US. It states that the creation or art must have a minimum level of creativity.

The Supreme Court of India, in the case of Eastern Book Company vs. D.B. Modak[4], moved from the sweat of the brow doctrine to the modicum of creativity doctrine. For AI generated art the latter is a better standard to determine the ownership of copyright. It might ensure that there is some minimum human intervention in the creation.

Copyright Infringement & AI

The problem in determining originality is that the generative AI models often feed on copyrighted content. Recently, Studio Ghibli art took the social media platforms by a storm, where people were using ai to make images which were the Studio Ghibli version. Studio Ghibli is an original Japanese animation style. AI developers could argue about fair and transformative usage but it does not hold water when the images are strikingly similar to the original. Section 14 of the Indian Copyright Act, 1957 states that the owner of copyright shall have the exclusive right to reproduce the work.  According to Section 51 of the act, copyright infringement occurs when a person performs an act which only the copyright owner is entitled to do.

The doctrine of fair use highlights that using a copyrighted material for educational, citing or learning purposes is not an infringement. The Indian Copyrights Act, 1957 provides a full list of acts not considered as copyright infringement under Section 52. However, using the original work to produce a substantially similar version of it, is an infringement. Generative AIs use copyrighted works to produce them in the same version. The expression of the idea is protected by the ownership of copyright and not the idea itself. The test to determine infringement is when the viewers or audience are incapable of determining which is the original work then it is a case of copyright infringement[5].

The question of liability persists here as well. The user or prompt giver rarely has any control over the data that is used by the AI to functions. If the developer knowingly engineers the generative AI to use copyrighted material to reproduce it, the developer can be held liable.

Cybercrime Using AI

Statistics show a hike in cybercrime committed using deepfake and other AI tools[6]. Recently, a member of parliament in New Zealand showed a deepfake nude picture of herself in the Parliament. Various crimes like identity theft, cheating, extortion, etc. can be easily committed using AI. The Information Technology Act, 2000 entails punishment for offences committed through the internet. However, it does not explicitly recognize artificial intelligence as a means to commit offences.

Here, also exists a lack of clarity regarding the liability in such cases. The prompt giver using the AI model to generate criminally chargeable or offensive content would definitely be liable but the mystery is whether the developer should be held liable or not. Does Section 79 of Information Technology Act, 2000, the ‘safe harbour clause’ protect AI developers as well? The developers must follow certain precautionary measures to ensure that their models are not exploited in manner which can have criminal repercussions. An important criterion for availing safe harbour clause is that the platform or the intermediary functions in a passive way and merely provides services. The intermediary must not actively engage in the act itself. AI models are of various kinds and the entire realm is a new one. One cannot have a set understanding of how AI models function only by assessing a few of them. An AI model can be curated to be a passive or an active performer. This can only be analysed on a case-to-case basis.

Are the pre-exiting laws enough?

AI is not a temporary phenomenon that will fade away eventually. In fact, it is here to stay. Currently the statutes available in India do not recognise AI as a means of committing offences. The pre-existing laws do not cater to the complexities that are involved in the matters of AI and are outdated. India is a member of the Berne Convention for the Protection of Literary and Artistic Works since 1928. The Berne Convention was adopted in 1886. It highlights the basic rights that the author of an original work must have across all the member countries. It is an international treaty that protects the intellectual property rights.

Countries like UK which had eased regulations to encourage the development of AI do not have a framework which regulates the working of AI. Policy is also needed to supervise AI developers and licensing for the same could be a step in the right direction. The European Union is formulating regulations for AI. The consequences and the ways in which artificial intelligence can be used is wide ranging. Therefore, regulations dealing with liability, copyright infringement and ownership, cybercrimes, etc. are the need of the hour. The European Union is formulating regulations for AI.

Conclusion

As AI continues reshaping the creative arenas, legal systems must adapt in time. The Indian Copyright Act, 1957 is outdated and needs an amendment which is comprehensive enough to address the intricacies of generative AI. The legal vacuum surrounding AI must not become a loophole for people to exploit it. AI is a dynamic and rapidly evolving subject. This makes it difficult for the legislatures to keep up and makes it prone to exploitation.

[1] Danielle S. Allen, ‘Is Art Generated by Artificial Intelligence Real Art?’ (Harvard Gazette, 17 August 2023) https://news.harvard.edu/gazette/story/2023/08/is-art-generated-by-artificial-intelligence-real-art/ accessed 7 June 2025.

[2] University of London Press Ltd v University Tutorial Press Ltd [1916] 2 Ch 601 (Ch).

[3] Feist Publications Inc v Rural Telephone Service Co 499 US 340 (1991).

[4] Eastern Book Company v DB Modak (2008) 1 SCC 1.

[5] RG Anand v Deluxe Films (1978) 4 SCC 118.

[6] Regula, The Impact of Deepfakes on IDV: A Regula Survey (Regula Forensics Blog, 2024) https://regulaforensics.com/blog/impact-of-deepfakes-on-idv-regula-survey/ accessed 7 June 2025.

-Eiha Verma