The Significance Of The ANI vs OpenAI Case

The Delhi High Court is hearing a case that will determine whether artificial intelligence companies can use copyrighted work to train their models without permission. Asian News International, a major Indian news agency, sued OpenAI in late 2024 for using its copyrighted news articles to train ChatGPT without authorization or compensation. The case has moved through 2025 and continues into 2026, and its implications extend far beyond India.

The case makes us think whether using content without permission for machine learning counts as copyright infringement. The answer seems obvious until you realize that the Indian Copyright Act was written in 1957, well before anyone imagined machines learning statistical patterns from text. Courts are being asked to interpret old law against new technology, and their decision will reshape how AI companies source training data and how creators protect their intellectual property globally.

ANI discovered the problem when ChatGPT generated responses containing verbatim excerpts from ANI articles. In some cases, ChatGPT fabricated interviews and falsely attributed them to ANI, potentially damaging the agency's reputation and editorial credibility. ANI requested a licensing agreement with OpenAI in October 2024. OpenAI declined. ANI sued, seeking ₹20 million in damages and an injunction preventing OpenAI from storing, publishing, or reproducing ANI's work (Lukmaan, 2026).

The core legal argument from ANI is that copying content without permission is infringement, regardless of purpose. The Copyright Act explicitly protects original literary work. Using that work to train a commercial AI system is not a permitted exception. It is copying and infringement.

OpenAI's defense shows that the two sides understand "training" completely differently. OpenAI contends that its model does not copy ANI's articles. The model learns statistical patterns and linguistic relationships from the training data. The final model contains no verbatim copy of ANI's work. What was copied during training is not stored in the model itself. Therefore, no infringement occurs.

OpenAI also challenges the jurisdiction. The company has no offices or servers in India. Its training operations occur in the United States. OpenAI argues that Indian courts cannot regulate conduct that happens outside Indian territory. But more importantly, OpenAI invokes fair dealing, arguing that using publicly available data for research and transformation falls under exceptions in India's Copyright Act.

But the Indian law diverges sharply from US precedent. Fair use in the United States is flexible and context-dependent. Courts weigh factors such as purpose, nature of the work, how much was used, and market effect. This flexibility allowed US courts to previously rule that Microsoft could use thousands of copyrighted books to train AI systems. The courts found this transformative and therefore fair use.

India's fair dealing exception is different and more restrictive. Section 52 of the Indian Copyright Act provides an exhaustive list of permitted uses. These include private research, criticism, comment, news reporting, teaching, and scholarship. Before you can even ask whether a use is fair, you must first establish that it falls within one of these enumerated categories. Copying thousands of news articles into a machine learning pipeline for a commercial product does not clearly fit any of these purposes. It is not news reporting or criticism or private research. It is industrial-scale data mining for commercial profit.

The court appointed two copyright experts to navigate this question, and they both disagreed. One expert argued that copying for storage falls under a permissible exception so long as nothing expressive is made public. The other argued that OpenAI's unlicensed copying for commercial use is infringement regardless of how the model stores or uses the data (LawSchoolPolicyReview, 2025).

This case holds importance because Indian courts are being asked to decide first. The New York Times sued OpenAI in the United States on similar grounds in 2023, but that case is still ongoing. US courts have not yet ruled definitively on whether fair use applies to AI training. Unlike the EU, which is writing explicit legislative exceptions for text and data mining, India has left courts to interpret 1957-era copyright law against 2020s-era technology. Whatever the Delhi High Court decides will establish precedent that ripples across the Global South, where legal systems share more similarities with India than with the United States or Europe.

The case has already attracted intervention from the Digital News Publishers Association, representing India's major media organizations, and the Indian music industry has also filed to intervene, anticipating that the ruling could affect AI training on music and other creative works. Everyone understands the stakes. If ANI wins, it signals that content creators can enforce copyright against AI training. Expect lawsuits from Indian publishers, music labels, and filmmakers. AI companies may need to license training data explicitly or face litigation. If OpenAI wins, it validates the "publicly available data" argument globally. The ruling would suggest that creators cannot prevent AI training on their publicly available work without explicit contractual restrictions.

Neither outcome settles the global question. US and EU courts will still interpret their own laws differently. But India's ruling will become one more data point in a rapidly fragmenting global landscape where copyright jurisprudence diverges by jurisdiction. For AI companies operating globally, this fragmentation means building jurisdiction-specific compliance frameworks. For creators, it would mean understanding which laws protect your work and which do not. And for the industry, it means that the same training data that is legal in one country may be illegal in another.