ANI appeals Delhi HC order denying interim relief against OpenAI

MUMBAI: The copyright battle is moving up a court tier, with news agency Asian News International (ANI) taking its fight with OpenAI to the Delhi High Court’s Division Bench. ANI has challenged the July 24 order that denied it interim relief in its copyright infringement case, putting one of India’s closely watched AI and copyright disputes back in the courtroom.

The appeal is listed for hearing on September 8 before Justices V Kameswaro Rao and Manmeet Pritam Singh Arora.

ANI, the first Indian media house to sue OpenAI over alleged copyright violation, had sought an interim injunction preventing the AI company from storing, publishing, reproducing or using its copyrighted works. Justice Amit Bansal, however, declined the request at the interim stage, holding that OpenAI’s storage of ANI’s original works for training the large language models behind ChatGPT fell within the fair-dealing exception under Section 52(1)(a) of the Copyright Act, 1957.

The single-judge ruling also found, on a prima facie basis, that ChatGPT responses were not substantially similar to ANI’s original literary works. It further said ANI had not established that its works had been memorised and reproduced through ChatGPT responses, leaving the agency without a prima facie case for an interim injunction.

ANI is now asking the Division Bench to take a different view. Its argument is that making news content publicly available does not amount to permission for OpenAI to commercially exploit, reproduce or store that material for AI training without authorisation. The agency has also challenged the single judge’s interpretation of fair dealing, arguing that it could stretch the exception for commercial purposes beyond what the Copyright Act permits.

That makes the appeal bigger than a dispute between one news agency and one AI company. At stake is a question increasingly confronting courts and publishers worldwide: when copyrighted material is copied and stored to train a generative AI model, is that part of the learning process protected by copyright exceptions or does it require permission from the rights holder?

The July 24 ruling was the first decision by an Indian constitutional court to examine copyright law and the fair-dealing exception specifically in the context of AI training and large language models. The court noted that AI training can serve purposes including scientific and technological research, dissemination of information, education and accessibility, while also weighing the interests of copyright owners.

ANI’s original suit was filed in the Delhi High Court in 2024. The agency alleged that OpenAI had used its copyrighted material without authorisation to train its models and raised concerns about its works being reproduced in ChatGPT responses. The case has also drawn the attention of other news industry stakeholders, including the Digital News Publishers Association, whose members include several major Indian media organisations.

For now, the single judge’s ruling remains confined to the interim application and does not decide the underlying suit. The court had also cautioned that granting an injunction at this stage could cause irreparable injury not only to OpenAI but potentially to the public interest.

The Division Bench will now consider whether the July 24 order warrants interference at the appellate stage. Whatever the outcome, the case is steadily pushing an old copyright rule into a very new-age courtroom question: how much can an AI learn from the news before the law says it has learned too much?