Delhi High Court seeks OpenAI response in ANI copyright appeal

MUMBAI: The copyright battle between ANI and OpenAI has entered another round, with the Delhi High Court now asking the ChatGPT maker to put its case on paper. A Division Bench of the Delhi High Court on Tuesday issued notice to OpenAI on an appeal filed by Asian News International (ANI) challenging an earlier order that refused to grant the news agency interim protection in its copyright dispute with the AI company.

The Bench, comprising Justices Avneesh Jhingan and Manmeet Pritam Singh Arora, sought OpenAI’s response and listed the matter for hearing in December.

ANI, which became the first Indian media organisation to sue OpenAI over alleged copyright infringement, filed the original case in 2024. The agency alleged that OpenAI had used its news content without authorisation to train and operate ChatGPT. It also claimed that ChatGPT could reproduce ANI material in response to user queries and, in some instances, attribute statements or events to ANI that had never occurred.

The dispute goes to the heart of a much bigger question for publishers and AI companies: where does copyright end and AI training begin?

In July, Justice Amit Bansal declined ANI’s request for an interim injunction. The single judge held that OpenAI’s storage of ANI’s works for training ChatGPT did not, at the interim stage, appear to amount to copyright infringement. The court also said granting interim relief could cause irreparable injury to OpenAI and have wider public-interest implications.

ANI has now taken that finding to the Division Bench, challenging the interpretation of Section 52 of the Copyright Act, 1957. The agency argues that Indian law provides for ‘fair dealing’, rather than the broader ‘fair use’ doctrine, and that statutory exceptions are restricted to specified purposes.

ANI has also questioned whether commercial use of copyrighted material on a large scale can fall within the exception relating to private or personal use, including research.

The agency further argues that OpenAI’s position that ANI material accounts for only a small portion of its training data does not settle whether that use is legally permissible. ANI has also pointed to OpenAI’s licensing deals with publishers, including the Financial Times, arguing that the existence of a market for licensed news content is relevant to the copyright question.

During Tuesday’s hearing, ANI counsel Sidhant Kumar referred to an interim arrangement under which OpenAI had agreed not to scrape content from ANI’s website. Kumar said the arrangement remained in place from September 11, 2024 until the single judge passed the July order.

The Division Bench, however, indicated that it would not issue directions on that arrangement before hearing OpenAI’s response.

Broadband India Forum, which had intervened before the single judge, also sought permission to assist the Division Bench. Senior Advocate Kapil Sibal, appearing for the forum, told the court that the dispute had wider implications for the technology and publishing industries.

OpenAI had previously challenged the maintainability of ANI’s case in India, arguing that its AI models were not trained in the country and its servers were located in the US. It also maintained that machine learning involves transformative use of material and that factual news itself cannot be protected by copyright.

The company had further said publishers could request that their websites be placed on a blocklist to prevent crawling.

The single judge had held that storing ANI’s original works for AI training did not amount to copyright infringement at the interim stage, viewing the training of large language models as a machine-learning process involving the screening and organisation of stored literary works.

The Division Bench’s notice does not decide the copyright dispute itself. It simply opens the next chapter, with OpenAI now required to respond before the court considers ANI’s challenge to the refusal of interim relief in December.