Delhi High Court Rules OpenAI’s Use of ANI Content to Train ChatGPT Is Not Copyright Infringement

In what is being hailed as India’s first substantive judicial pronouncement on AI training and copyright law, Justice Amit Bansal of the Delhi High Court on July 24, 2026, refused to grant news agency ANI (Asian News International) an interim injunction against OpenAI. The Court held that OpenAI’s storage and use of ANI’s copyrighted news articles to train ChatGPT does not, at the prima facie stage, amount to copyright infringement. The 135-page judgment lays down an analytical framework for applying the fair-dealing exception under Section 52 of the Copyright Act, 1957 to artificial intelligence systems, marking a landmark moment in Indian intellectual property law.

Background of the Case

ANI, one of India’s largest news agencies, had filed a copyright infringement suit against OpenAI, alleging that the company scraped and stored its copyrighted news articles to train its large language model, ChatGPT. ANI contended that this constituted unauthorised reproduction of its literary works and sought an interim injunction to restrain OpenAI from using its content for training purposes. The case has been closely watched by the technology industry, media houses, and intellectual property practitioners across India, as it represents the first time an Indian court has been asked to rule on whether AI companies can use copyrighted works to train foundation models.

The Court’s Analysis on Fair Dealing Under Section 52

The central question before the Court was whether OpenAI’s use of ANI’s copyrighted content fell within the scope of the fair-dealing exception under Section 52(1)(a)(i) of the Copyright Act, 1957. This provision permits fair dealing with a literary, dramatic, musical, or artistic work (other than a computer programme) for the purposes of private or personal use, including research.

Justice Bansal held that, prima facie, OpenAI’s use of ANI’s content to train its large language models was permissible and fell within the scope of private or personal use, including research. The Court observed that the training process involves computational analysis of large datasets to identify patterns and statistical relationships within the text, rather than copying the content for direct publication or distribution. This distinction was critical in determining whether the use constituted fair dealing.

The judgment also examined the transformative nature of the use. The Court acknowledged that the output generated by ChatGPT is not a reproduction or copy of the original copyrighted works but rather a synthesis of patterns learned from vast amounts of data. This transformative quality weighed in favour of OpenAI’s fair-dealing defence.

The Question of Substantial Similarity

A second significant finding related to the question of substantial similarity between ChatGPT’s outputs and ANI’s original works. ANI had argued that ChatGPT’s responses sometimes contained information that closely mirrored its reporting. However, Justice Bansal noted that ANI had failed to satisfy the Court that memorisation and regurgitation of the agency’s original literary works had occurred through the responses generated by ChatGPT.

The Court observed that factual information, once published, does not enjoy copyright protection; what is protected is the original expression of that information. Since ANI could not demonstrate that ChatGPT’s outputs reproduced the specific expression of its articles, the claim of substantial similarity did not hold at the interim stage.

Implications for AI Companies Operating in India

This ruling carries significant implications for AI companies operating in India. While the judgment is limited to the interim stage and the main suit remains pending, the Court’s analytical framework provides considerable clarity on how Indian courts may approach the intersection of AI training and copyright law. The fair-dealing defence under Section 52 has historically been applied in traditional contexts such as academic research, criticism, and news reporting. By extending its application to computational analysis by AI systems, the Delhi High Court has signalled that Indian copyright law may accommodate new technological uses of copyrighted works, provided the use is genuinely transformative and does not result in the reproduction of protected expression.

For media organisations and content creators, the ruling underscores the importance of demonstrating actual harm to copyrighted expression, rather than relying on the mere fact that content was used in training datasets. The Court’s emphasis on the absence of memorisation and regurgitation suggests that future plaintiffs will need to present concrete evidence that AI outputs reproduce the specific creative expression of their works. Recent Delhi High Court rulings on trademark protection demonstrate that Indian courts continue to take intellectual property rights seriously, but they also require claimants to meet the evidentiary threshold.

The Broader Context of AI and Copyright Disputes

India is not alone in grappling with these questions. Courts in the United States and the European Union have been dealing with similar disputes, including high-profile cases filed by The New York Times against OpenAI and by Getty Images against Stability AI. The Delhi High Court’s ruling places India among the first jurisdictions to issue a substantive judicial finding on AI training and copyright, and its reasoning is likely to be cited in ongoing debates over proposed amendments to copyright law in various countries.

The judgment also has implications for India’s evolving data governance framework. The interplay between copyright law, fair dealing, and the Digital Personal Data Protection Act remains an evolving area. As AI companies increasingly rely on publicly available content for training purposes, the boundaries between intellectual property protection and data privacy rights will require further judicial and legislative clarity. Questions about the right to be forgotten in the context of AI-trained models add further complexity to this debate.

What Happens Next

The interim order does not dispose of the main suit, which remains pending before the Delhi High Court. ANI may challenge the interim findings during the trial, and the final judgment could take a different view on the merits. As the matter involves significant questions of law affecting the technology and media industries, an appeal to the Division Bench or the Supreme Court remains a possibility.

Key Takeaways

1. The Delhi High Court held that OpenAI’s use of ANI’s copyrighted news content to train ChatGPT prima facie constitutes fair dealing under Section 52(1)(a)(i) of the Copyright Act, 1957.

2. The Court found that AI training involves computational analysis of patterns, not direct reproduction or distribution of copyrighted content.

3. ANI failed to demonstrate that ChatGPT memorised and regurgitated its original literary expression.

4. This is the first substantive Indian court ruling on whether AI training on copyrighted content constitutes infringement.

5. The main suit remains pending, and the final outcome could differ from the interim findings.