Copyright In ChatGPT

The purpose of OpenAI is to "guarantee that artificial general intelligence (AGI) benefits all of humanity," according to its charter. The three categories of weak, strong, and superintelligence have been used to categorise AI systems in order to provide clarification. Siri or Alexa are examples of weak or narrow AI because they are only capable of doing one task or finding one solution. Strong AI, also known as artificial general intelligence, is AI that has human-like intelligence in terms of innovation and creativity, including the ability to learn and improve abilities while doing various tasks. Superintelligence is a type of intelligence that, to put it simply, would excel in every sector beyond that of the smartest humans.

According to the Terms of Use (ToU), "as between the parties and to the extent permitted by applicable law, you own all Input. Subject to your compliance with this Agreement, OpenAI now assigns to you all its right, title, and interest in and to Output." The user's inquiry is referred to as the input, while the chatbot's response is referred to as the output. On the basis of the existing legal climate, the portion of the ToU mentioned above won't be enforceable in the US or India.

The copyright ramifications of ChatGPT can be conveniently split into two categories: input and output. The discussion of the potential effects of the output has only been conducted under the presumption that a human being provided the input to the question, and that if the requirements for copyrightability are satisfied, the human being would be given authorship and ownership (or not). When it comes to the Output, the interesting queries include if it complies with copyright protection standards.

It appears insane to allocate output to the user and would be unsuccessful on many levels, both technically and practically. Let's assess the technological problems from an Indian perspective. First off, it is not legally tenable to assume that ChatGPT immediately transfers ownership of any work it creates to OpenAI. Can AI create the output it produces on its own, or is the creator of the AI or the person providing the input more likely? According to the ToU, we would need to look at Section 17 of the 1957 Copyright Act in order to determine whether OpenAI is the rightful proprietor of ChatGPT's output (the Act). Nonetheless, India is required to have human authors under Section 17, as shown by: Sec. 16 of the Act states that "no person" is entitled to copyright except as provided by law; Sec. 2(d)(vi) of the Act permits grant of authorship to the person "who causes the work to be created" in the case of computer-generated works; and Sec. XIV, Application for Registration of Copyright, requires disclosure of name, nationality, and address. In accordance with Section 2(d), a human who has only contributed a single-line Input cannot be considered to have caused the Output to be formed (vi). Consequently, it would appear that no one is permitted under Indian law to claim authorship (or co-authorship or ownership) of an AI-generated work. Interestingly, as per the online records of the Copyright Office, 'Raghav Artificial Intelligence Painting App' (AI) still continues to be registered as a co-author (along with a human being), despite the issuance of a withdrawal notice more than a year ago.

After the aforementioned barrier has been overcome, how will the user demonstrate that it has been lawfully granted copyright for the Output? By omitting certain information necessary by Section 19 of the Act, such as term, territory, and the amount of any royalty or other payment due to the author, the ToU fails to meet the conditions for a legitimate assignment. For an artwork or piece to be published online, the region should ideally be the entire world. It would also be challenging to practically remove the same after 5 years. One item might possibly be removed, but how would this work when there are dozens or perhaps millions of such works? A work can be exploited even in the absence of specification of term and territory (albeit to a limited extent), but an assignment under Indian copyright law cannot be made without stipulating the amount of royalty and consideration payable to the author.

In addition to the aforementioned technical obstacles, the ToU acknowledges that the Output "may not be unique across users" for inquiries of a similar nature. Consider a situation where one person asserts copyright over a certain output while a second user asserts copyright over the exact same output that he also created independently. Now picture these kinds of assertions occurring frequently. As no two persons can create the same exact play or book, the concept of honest and continuous usage does not apply to copyright law, unlike trademark law. The outcome would be unpredictable in these circumstances, unlike ordinary copyright infringement lawsuits when one party is typically clearly at fault. This would mean that such Output cannot be the subject of copyright protection.

According to the ToU, the user is in charge of the input and output, "including for ensuring that it complies with all applicable laws" and the ToU. What happens when someone's copyright has been accidentally violated even though the Work is not covered by copyright protection? When questioned, ChatGPT admits that it was trained using a significant amount of text data. Yet, ChatGPT is not given credit for creating the Output in the first place. If permission has been gained for such data, it is unclear. This puts the user who asserts copyright over the Output in a challenging situation because the use might not even be considered to be fair dealing. When the defendant websites/OTT platforms in infringement lawsuits before the Calcutta High Court and Delhi High Court permitted streaming of sound recordings protected by copyright law and claimed exception under Section 52(1)(a)(i) of the Act, 1957 (i.e., private or personal use), the Calcutta HC found that the streaming of sound recordings (protected by copyright law) for a "fee or revenue may be from its sponsors or from third parties" was commercial exploitation, and the defendant


The user's prompts and completions may be posted to social media and live streaming under the terms of OpenAI's Sharing & Publishing Policy, subject to a few conditions, one of which is to "note that the content is AI-generated in a way no user could reasonably miss or misunderstand." This appears to be a clickwrap agreement because OpenAI indicates during the registration process that by clicking "Continue," the user accepts its Terms. The Terms are not explicitly presented to or accepted by users when they choose to sign up using Gmail. Although the ToU states that all rights are assigned, it appears that the AI somehow still has the right to paternity. From one perspective, this is definitely a good thing to prevent human authors from trying to pass the Output off as their own work. While at the same time claiming credit for the Output, they fail to acknowledge their source or even disclose the scope of their training data.

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.