Embracing AI, copyright intact

One of the vexing questions confronting courts and lawmakers today is how to ensure that the law keeps pace with the rapid emergence of artificial intelligence (AI) solutions. Intellectual property (IP) law is no exception to this trend. For instance, courts are being confronted with complex issues such as determining the inventor of a patented invention or the author of a copyrighted work created by AI.

A pertinent case currently before the Delhi High Court involves Asian News International (ANI) suing OpenAI, the creator of ChatGPT, in November 2024. ANI alleges unauthorised use of its copyrighted material for AI training purposes. ANI contends that OpenAI has incorporated its news content into ChatGPT’s training data without obtaining the necessary licenses, thereby infringing on ANI’s copyright. Additionally, ANI asserts that ChatGPT has, on occasion, generated fabricated quotations or statements erroneously attributed to ANI, a phenomenon known as AI hallucination. The Delhi HC has framed four critical questions to address in this lawsuit:

Storage and Copyright Infringement: Does OpenAI’s storage of ANI’s data, which is protected under the Copyright Act, 1957, for training ChatGPT constitute copyright infringement? Use in Response Generation: Does OpenAI’s use of ANI’s copyrighted data to generate responses for ChatGPT users amount to copyright infringement? Fair Use Defence: Can OpenAI’s use of ANI’s copyrighted data be considered fair use – a defence that allows the use of copyrighted works without the copyright owner’s consent in some well-defined circumstances? Jurisdictional Authority: Do Indian courts have jurisdiction over this lawsuit, given that OpenAI’s servers are located in the United States?

On Friday (February 21), the HC heard submissions from two amicus curiae (friends of the court) appointed by it, as it grapples with this novel legal issue. Amicus curiae Arul George Scaria, a professor of IP law at the National Law School of India University, Bengaluru, presented his submissions. Scaria argued that the court possesses jurisdiction in this matter, emphasising that ChatGPT is accessible and used within India. This perspective suggests that the location of servers is less significant than the locus of content creation and consumption.

Regarding the alleged infringement, Scaria acknowledged that OpenAI’s use of ANI’s materials would constitute reproduction of copyrighted work. However, he contended that it could potentially fall under the ‘fair dealing’ exceptions, depending on the specific facts of each case. To strike a balance between the rights of copyright owners and public interest, the law creates a set of ‘fair dealing’ exceptions. These exceptions allow for the use of copyrighted works for purposes such as criticism, quotation, and parody without having to obtain the copyright owner’s prior approval.

One key question is whether AI developers must obtain a licence from copyright owners before using their work. Scaria argued that an AI developer like OpenAI has deep pockets and, therefore, can afford licensing fees. However, this licensing requirement would become a barrier for small AI developers and impede innovation. Scaria highlighted that restricting AI’s access to diverse materials might inadvertently increase the risk of misinformation, as AI models rely on extensive datasets to generate accurate and reliable outputs.

Another amicus, advocate Adarsh Ramanujan, argued that policy-based considerations should be kept out of the calculus and the case can be decided within the Copyright Act. The differences in approaches between the two amici could be attributable to the different functions they serve – Scaria is a thoughtful academic and Ramanujan a practitioner of black letter law.

India is not the only country dealing with this issue; more than 30 cases are pending in different jurisdictions on this issue. It all began in 2023, with the New York Times suing OpenAI. The NYT argued that OpenAI allows users to access content identical to or substantially similar to that found on the NYT’s platform. This threatens its business model and journalistic integrity, it contended.

As the HC continues to hear this case, it will have to find a robust balance point between the need to ensure that the AI industry is allowed to flourish while ensuring that copyright owners do not get a raw deal by virtue of AI-induced unauthorised reproduction of their copyrighted material. While the role of the court is important, the issue may ultimately require a legislative solution to ensure that the competing considerations at play are balanced in a thoughtful and futuristic manner. As we await the court’s verdict and potential law reform, it is imperative for stakeholders – including technologists, legal experts, policymakers, and content creators – to engage in collaborative discussions. Such dialogues are essential to develop frameworks that both encourage innovation and protect the rights of those who contribute original content, ensuring a harmonious integration of AI into our societal and legal fabric.