The development and use of ChatGPT have raised various questions about the data generated. This article argues that when there is a substantial intervention of the user in the form of prompts then AI might not generate a similar response for multiple users and hence the user shall be able to claim copyright. However, the current legal framework does not allow the work generated by AI to be claimed under copyright. Here, we discuss the copyright law and judicial framework and how the concept of copyright needs to be revisited given the change in technology.
INTRODUCTION
ChatGPT is an Open artificial intelligence tool that uses prompts to generate a response. It uses the method of conversation wherein users give a prompt in the form of encoding which is used to generate a response. This response is in natural human-friendly language.
Eg. Prompt: What is ChaptGPT?
Response: Chatbot GPT, or ChatGPT, refers to a conversational artificial intelligence (AI) system based on the GPT (Generative Pre-trained Transformer) architecture developed by OpenAI. GPT models are designed to generate human-like text responses based on the given input. They are trained on a vast amount of text data from the internet, allowing them to understand and generate coherent and contextually relevant responses in natural language.
ChatGPT is a specific variant of the GPT model that is fine-tuned and optimized for conversational interactions. It can be used in various applications, such as customer support, virtual assistants, or simply as a chatbot for general conversation. ChatGPT utilizes its training on large-scale text data to generate responses that simulate human-like conversation, providing users with interactive and engaging experiences.
OpenAI has released several versions of the GPT model, with GPT-3 being the most well-known prior to my training. However, please note that I am based on the GPT-3.5 architecture, and my responses are generated using the knowledge available up until September 2021.
ChatGPT uses various sources of information like articles, blogs, and papers to generate this response. The fundamental question which comes to mind is who owns the data generated by ChatGPT. In terms of Use, ChatGPT gives all rights, titles, and ownership of output to the person generating it. This brings us to the next question that is can ChatGPT transfer these rights to users when it doesn’t possess them in the very first place? The data used by ChatGPT is already existent- this data either lacks copyright or is similar in manyfold ways to existent information. Whether AI can create something new or original is still questionable. For understanding the basics of claiming copyright, let us first understand the copyright law in India.
LEGISLATIVE FRAMEWORK
Copyright is a legal entitlement given to the creator of original work. In India, it is governed by the Copyright Act, of 1957. Section 13[1] of the act specifies that copyright is given to the creator of the original literacy, dramatics, artistic and musical works. It extends to cinematograph movies and sound recording. Ideas cannot be protected but expression is protected.
Section 14[2] of the Act gives the owners of copyright certain exclusive rights such as publication, adaptation, reproduction etc. Section 17[3] of the Act talks about the first owner of work. Eg. Where a photograph was taken for valuable consideration at the instance of another then the person at whose instance the work was done would be considered as the first owner. Under the copyright law in India, the author and owner of the copyright can be two different people. The author will have moral rights like paternity and integrity. On the other hand, the author will have a right to monetarization, etc. The amendment of 1994, brought computer-generated work in literacy, art, dramatics, and music under Section 2(d)(v)[4] as ‘the person who caused the work to be done.’
When we read Sections 13 and 14 together, we understand that the litmus for copyright claim is original work. AI works on programming for generating results that explore and analyze already existing information. This brings us to our primary question can work generated by AI be considered original work? Let us look at the judicial framework in this regard.
JUDICIAL FRAMEWORK
The Hon’ble Delhi High Court, in Rupendra Kashyap v. Jiwan Publication House Pvt. Ltd[5]., in deciding the question related to the authorship of question papers and their compilation and whether the Central Board of Secondary Education can claim copyrights over question papers. It was held that CBSE was an artificial person and could not claim copyright unless it is established and proved that persons were engaged in the compilation and preparation of question paper. Thus, establishing that only a natural person can claim copyright in Indian jurisdiction. This view was upheld in Tech Media Private Ltd. v. Jyoti Janda[6], wherein it was held that authorship cannot lie with a Juristic person, though it may be a copyright owner.
In Navigator Logistics Ltd. v. Kashif Qureshi & Ors.[7], copyright claims were rejected over a list compiled by computed because of lack of human intervention.
ANALYSIS
The content on ChatGPT includes both the input and the output. The input is primarily owned by the user, but the output generated is in question here whether it is owned by the user or AI. There are legal and technical difficulties associated with the content generated by AI. Copyright law in India’s requirement is a human creation and some element of creative input. When we look at the consideration of the one who causes work to occur, it shall be substantiated by some human efforts. Now, when we talk about ChatGPT, it uses human prompts to generate any form of response or work. Therefore, it shall be treated as a tool to generate work. Another problem is that AI may generate similar data for two individuals then who could claim copyright over the same piece of work. However, if the prompts are unique, the probability of a similar response will be reduced. At the same point in time, simpler and abstract concepts are outside the purview of copyright claims.
Answering the question of copyright, whether it can be claimed by the person who developed AI technology, AI or the user. Giving the person who created AI claims over all the data generated over it, this in itself sounds abrupt, someone who just did the programming cannot be the owner of everything generated over it.
Giving AI ownership comes with it own sets of problems. This is similar to the Monkey selfie case of copyright[8]. It is just a tool for generating work and cannot address any infringement or jurisdictional issues.
The user is however the person who applies his mind to come up with a prompt on the basis of which work is generated. He must be treated as the owner of the copyright and AI must be seen as a tool for generating data as it is his application of mind that is facilitating the creation of work. Expressions are subjected to protection.
With the development of AI, it has become highly relevant to answer these questions especially when technology changes. When the copyright law came into existence the lawmakers could not have foreseen this development and therefore, the concept of copyright is very rudimentary. There is a need for lawmakers to revisit the copyright law to accommodate technological changes because AI can come, it is not going anywhere, and people will gradually adapt to it for more time-consuming and mundane tasks. In this situation, where a person is directing AI to work in a particular direction, giving intricate details such as writing style, should the work not be recognized only because of the use of AI is the question that needs to be answered.
CONCLUSION
Time and again we have seen with a change in circumstance, the law needs constant reformation. With the increase in cyber-related offenses, we realized that the one of primary possession of a person is his or her information and right to disperse that information, accordingly, we felt the need for a new law. Similarly, now is the time to start discussions and debates surrounding AI. It must be treated as a tool for generating work. The policymakers need to ponder under what circumstances human involvement in the creation of work by AI can be entitled to copyright claims. Privacy concerns surrounding AI also need to be addressed. Perhaps, if there is better clarity on the method used by ChatGPT, copyright claims can be better addressed.
[1] Section 13 in the Copyright Act, 1957
[2] Section 14 in the Copyright Act, 1957
[3] Section 17 in the Copyright Act, 1957
[4] Section 2(d) in the Copyright Act, 1957, Amendment 1994
[5] 1996 (38) DRJ 81
[6] (2014) 60 PTC 121
[7] 254 (2018) DLT 307
[8] Adres Guadamuz, Can monkey selfie teach us anything about copyright law? WIPO Magazine (Jun. 26, 2020, 6:10 PM), https://www.wipo.int/wipo_magazine/en/2018/01/article_0007.html
Shruti Vijayvargiya
Author
As a third-year law student at NMIMS’s Kirit P Mehta School of Law, I have developed a keen interest in intellectual property law and its intersection with emerging technologies.
I believe that the rapid advancements in artificial intelligence (AI) have posed significant challenges to the existing copyright framework. With AI systems becoming increasingly capable of generating creative works, it is imperative to reassess copyright laws to ensure adequate protection for both creators and consumers in this new digital landscape.
Leave a Reply