Akanksha Vishwakarma is a 3rd year student at Gujarat National law University Gandhinagar
Introduction
The use of generative artificial intelligence(AI) such as ChatGPT developed by OpenAI, Gemini by Google and Copilot by Microsoft, has gone significantly high since their introduction. The most recent data available on Reuters indicates that ChatGPT has around 180.5 million users[1] in which India stands as the second-highest number of people using the application. These AI use already available data on internet through “web scrapping” to collect data from various sources available on internet to provide output to the user. Questions have been raised regarding the reliability of the output generated, its implications on copyright and IP laws, ownership and authorship of the data, and liability for the data generated. It is also clear that trying to hold intellectual property created by AI and include it under the copyright framework presents a number of challenges.
Ownership of such content
The introduction of such websites has reinstated the issue of copyright and ownership over data provided by these software. The data produced by them is a recollection of the already copyrighted available data on internet. This raised question as to whether these generative AI chatbots can be considered as “authors or creators” of the original content for the purposes of various laws including copyright laws[2]. The authorship of the content remains questionable as to whether AI chatbots would be considered as author or the user who is using the chatbot. On being asked to “ChatGPT” itself on the question of ownership of its content, the answer was “The content generated by ChatGPT belongs to OpenAI, the organization that developed and operates the ChatGPT model. Users retain ownership of the input they provide to ChatGPT, but any output generated by the model is owned by OpenAI”.
It is a known fact that AI chatbots like ChatGPT, Gemini and Copilot uses already available data on the internet and surf through the data to provide output to the user[3]. Now the available data are already copyrighted and owned by someone else and thus when chatbots use their data to provide output to the user, it actually infringes upon the copyright of those organisations or websites. Thus while using the output of ChatGPT the user may unwillingly end up infringing upon someone else’s copyright.
The general principle underlining ownership in copyright is that the person who creates a work is the owner and is considered as the “first owner” of the copyright under Indian law as per Section 17 of TheCopyright Act,1957 which defines first owner of copyright as “the author of a work shall be the first owner of the copyright”[4]. So, in case of chatbots, they generate output via web surfing and therefore the first owner of the copyright in that output is those chatbots. But as we know the output generated by them is merely a paraphrased or summarised version of the already available copyrighted data on the internet, therefore it is an infringement of the copyright of the available data. Ownership of the content depends on the sources used in the training database, who owns that database, and how closely the generated output resembles the database. If ChatGPT is considered capable of producing original literary works, it could be seen as an author, which brings up potential copyright infringement issues. However, OpenAI’s terms suggest that liability for infringement shifts to the user who generates the output, as OpenAI assigns ownership of the generated content to the user who provided the input.
Challenge on intellectual property rights
The IP debate pertaining to artificial intelligence revolves around the discourse regarding the authorship and ownership of the output generated by chatbots like ChatGPT and Copilot. The first question to be addressed is regarding the management of IP rights pertaining to the data used in the training of these programs. The second question to be addressed is whether the use of copyrighted material used in the training is an infringement of the copyright of the original owner. Various methods such as generative deep learning techniques and text and data mining(TDM) are used to train the AI algorithms. Text and data mining involves analyzing vast amounts of already available copyrighted data on the internet. By analyzing these text and data, AI algorithms learn from the pattern and therefore makes predictions on the question asked by the user. A significant challenge emerges in the fact that unlike humans, AI algorithms require exact copy of the artwork in their training period, this involves the preparation of millions of copies of copyrighted images, texts and data. The question arises whether the use of such data sets would act as an exception under “fair use”.
It is uncertain how the law would apply in this case, since ChatGPT is not a human and cannot be held accountable for its actions in the same manner as a person. Under Indian law, an author is defined as “the person who causes the work to be created, in relation to computer-generated work”[5], under Section 2(d)(vi). Therefore as per the definition provided, the person causing the work to be created is actually the user because it is because of his intervention that the output is created. Thus when he asks a particular question to the Chatbot and the output generated infringes upon the copyright of the original author, the person who caused the work to be created, i.e. user is to be held liable on infringement. But on the other hand, one may argue that it is because of the Chatbot that such infringement occurred and the user without knowing the source of the output used that data “fairly”. However, it is clear that because of the lack of strict laws on infringement through generative AI, liability cannot be determined in a straight-jacket formula[6].
According to the terms of CHATGPT, it mentions that “you own all Input, and subject to your compliance with these Terms, OpenAI hereby assigns to you all its right, title and interest in and to Output”. The question as to whether AI Platform hold the IP rights is still debatable under Indian laws. So far as the Indian perspective is concerned we can hold that the platform cannot hold the IP rights of the output generated.
Conclusion
It is a well-accepted fact that the use of AI Chatbots has been found useful in various sectors, but it is equally important to consider the impact it may have on intellectual property rights and the probability of risks associated with it. The question of ownership and authorship still remains and current Indian laws lack in covering AI under the Indian Copyright Act. The emergence and popularity of ChatGPT, Gemini, Copilot and other similar chatbots raises pertinent IP concerns that need to be addressed at the earliest.
[1]https://www.reuters.com/technology/chatgpt-sets-record-fastest-growing-user-base-analyst-note-2023-02-01/.
[2] Lucchi, N. (2023) ‘ChatGPT: A Case Study on Copyright Challenges for Generative Artificial Intelligence Systems’, European Journal of Risk Regulation, pp. 1–23. doi:10.1017/err.2023.59. https://www.cambridge.org/core/journals/european-journal-of-risk-regulation/article/chatgpt-a-case-study-on-copyright-challenges-for-generative-artificial-intelligence.systems/CEDCE34DED599CC4EB201289BB161965
[3] Varsha Jhavar, ‘ChatGPT and the Underlying Copyright Malady’, Spicy I: De-Coding Indian Intellectual Property Law. https://spicyip.com/2023/03/ai-and-copyright-law-analysing-the-impact-of-chatgpt.html.
[4] The Copyright Act, 1957, § 17, No.14, Acts of Parliament,1957.
[5] The Copyright Act, 1957, § 2(d), No.14, Acts of Parliament,1957.
[6] Ron Dreben, Mattthew T. Julyan, Generative Artificial Intelligence And Copyright Current Issues,Morgan Lewis, March 23,2023. https://www.morganlewis.com/pubs/2023/03/generative-artificial-intelligence-and-copyright-current-issues.