OpenAI can use copyrighted material to train ChatGPT, says Delhi High Court

News agency ANI had moved the Delhi High Court, accusing OpenAI of using its copyrighted material to train its large language models.

4 min readNew DelhiUpdated: Jul 24, 2026 04:10 PM IST

Every subsequent decision by that ministry acquires a shareholder value dimension, whether or not anyone intends it.

The Delhi High Court ruled on Friday that OpenAI training its large language models (LLMs) on original retrieved copyrighted works constitutes fair dealing and does not amount to copyright infringement.

News agency ANI Media Pvt Ltd had moved the Delhi High Court in 2024 against Open AI Inc (OAI) and Open AI OpCo LLC, raising issues of infringement of its copyrighted material for LLM training by OpenAI.

The observation on fair use of copyrighted material is a first for an Indian constitutional court.

“I am of the prima facie of the view that OpenAI’s act of storing ANI’s original retrieved works for training LLMs underlying ChatGPT falls under Section 52(1)(a) of the Copyright Act (fair dealing) and therefore does not amount to infringement…,” Justice Amit Bansal said on Friday, pronouncing his ruling in the open court.

Explained | Who are the architects of AI, and what are the ties that bind them?

“I am also of the prima facie view that outputs generated by ChatGPT using RAG (retrieval-augmented generation) technique does not amount to infringement under the Copyright Act since the outputs generated by OpenAI were not substantially similar to ANI (inputs),” Justice Bansal added.

ANI, he held, had failed to satisfy the court that “any memorisation or regurgitation of ANI’s original retrieved works has happened through the responses generated by ChatGPT”.

“In view of the discussion above, ANI has failed to make a prima facie case for interim injunction,” the court concluded.

Story continues below this ad

Further observing that the balance of convenience lies in OpenAI’s favour, the court reasoned that “irreparable injury would be caused not only to OpenAI but also to the public at large” if any interim injunction was granted in ANI’s favour.

OpenAI is an American artificial intelligence research organisation headquartered in San Francisco, which has developed the ChatGPT series of generative AI models.

Also Read | ‘Catastrophic’: Supreme Court warns against unregulated AI use in rulings

Not storing data directly: What OpenAI said

ANI had opposed OpenAI’s use of its publicly available copyrighted material to train its LLMs, verbatim replication of copyrighted material when a query is put in ChatGPT, as well as hallucinatory responses being provided by ChatGPT, while falsely attributing it to ANI.

OpenAI had submitted that it was not storing data directly and that there was no general prohibition on the use of data under copyright law–akin to the way in which reading a book could not be equated with “using it”.

Story continues below this ad

OpenAI had also argued that the pre-training process took place outside India, that training data was stored on foreign servers, and that OpenAI, over time, had refined its models to prevent regurgitation. It also said that following the pre-training phase, OpenAI’s models did not have access to the original training data after a cut-off date.

In November 2024, OpenAI had told the court that they had already “blocklisted” ANI’s domain, which meant that it was excluded from future training of OpenAI’s software.

Several other organisations, including media and publishing organisations and associations such as the Federation of Indian Publishers, Digital News Publishers Association, and the Indian Music Industry, who had raised similar objections to OpenAI’s training model, were parties in the suit.

Sohini Ghosh

Sohini Ghosh is a Senior Correspondent at The Indian Express. Previously based in Ahmedabad covering Gujarat, she recently moved to the New Delhi bureau, where she primarily covers legal developments at the Delhi High Court Professional Profile Background: An alumna of the Asian College of Journalism (ACJ), she previously worked with ET NOW before joining The Indian Express. Core Beats: Her reporting is currently centered on the Delhi High Court, with a focus on high-profile constitutional disputes, disputes over intellectual property, criminal and civil cases, issues of human rights and regulatory law (especially in the areas of technology and healthcare). Earlier Specialty: In Gujarat, she was known for her rigorous coverage in the beats of crime, law and policy, and social justice issues, including the 2002 riot cases, 2008 serial bomb blast case, 2016 flogging of Dalits in Una, among others. She has extensively covered health in the state, including being part of the team that revealed the segregation of wards at the state’s largest government hospital on lines of faith in April 2020. With Ahmedabad being a UNESCO heritage city, she has widely covered urban development and heritage issues, including the redevelopment of the Sabarmati Ashram Recent Notable Articles (Late 2025) Her recent reporting from the Delhi High Court covers major political, constitutional, corporate, and public-interest legal battles: High-Profile Case Coverage She has extensively covered the various legal battles - including for compensation under the aegis of North East Delhi Riots Claims Commission - pertaining to the 2020 northeast Delhi riots, as well as 1984 anti-Sikh riots. She has also led coverage at the intersection of technology and governance, and its impact on the citizenry, from, and beyond courtrooms — such as the government’s stakeholder consultations for framing AI-Deepfake policy. Signature Style Sohini is recognized for her sustained reporting from courtrooms and beyond. She specialises in breaking down dense legal arguments to make legalese accessible for readers. Her transition from Gujarat to Delhi has seen her expand her coverage on regulatory, corporate and intellectual property law, while maintaining a strong commitment to human rights and lacuna in the criminal justice system. X (Twitter): @thanda_ghosh ... Read More

Loading Recommendations...

Aug 11: Latest News

Live Blog