The Delhi High Court on Friday refused to grant interim relief to Asian News International (ANI) in its copyright infringement suit against OpenAI, holding that the storage and use of ANI’s copyrighted news content for training ChatGPT’s Large Language Models (LLMs) at the prima facie stage was protected under the fair dealing exception contained in Section 52(1)(a) of the Copyright Act, 1957, and therefore did not amount to copyright infringement.
Pronouncing the interim order, the single-judge Bench of Justice Amit Bansal first held that the High Court had territorial jurisdiction to entertain the suit despite OpenAI’s contention that its AI models were trained and hosted on servers located in the United States.
On the merits of the interim injunction application, the Court observed that OpenAI’s act of storing ANI’s original literary works for training its LLMs amounts to a non-expressive and transformative use, which is prima facie protected under Section 52(1)(a) of the Copyright Act and consequently does not constitute infringement under Section 51 of the Act.
The Court further held that responses generated by ChatGPT, including those generated through the Retrieval-Augmented Generation (RAG) framework, do not infringe ANI’s copyright as the AI-generated outputs are not substantially similar to ANI’s original news reports. It observed that ANI had failed to establish that ChatGPT had memorised, regurgitated or verbatim reproduced its copyrighted literary works while responding to users’ queries.
Holding that ANI had failed to satisfy the triple test governing the grant of interim injunctions, namely the existence of a prima facie case, balance of convenience and irreparable injury, the Court declined to restrain OpenAI from using ANI’s content during the pendency of the suit.
It observed that granting an interim injunction at this stage would cause irreparable injury not only to OpenAI but also to the public at large by adversely affecting access to and development of artificial intelligence technology. Accordingly, the Court dismissed ANI’s application for interim relief. The detailed judgment is awaited.
ANI had instituted the suit in 2024, becoming the first Indian media organisation to sue OpenAI for alleged copyright infringement. The news agency alleged that OpenAI had unauthorisedly copied, stored and commercially exploited its copyrighted news reports to train and operate ChatGPT.
It further alleged that ChatGPT, in certain instances, reproduced ANI’s reports verbatim or in substantially similar form and generated fabricated responses falsely attributing statements and news reports to ANI. According to ANI, such AI “hallucinations” posed a serious threat to its reputation, facilitated the spread of misinformation and fake news, and had the potential to undermine public confidence.
OpenAI opposed the interim relief, contending that copyright protection does not extend to facts or information contained in news reports and that the machine learning process involved in training LLMs is transformative in nature. It further argued that publishers unwilling to permit their content to be crawled or used for training could opt out by placing their websites on OpenAI’s blocklist. It informed the Court that ANI’s website had already been excluded from future training datasets.
The proceedings subsequently witnessed intervention applications from several media organisations and industry bodies, including the Digital News Publishers Association (DNPA), whose members comprise leading Indian media houses such as The Times Group, Hindustan Times, India Today Group and The Hindu.
In November 2024, the High Court framed important questions for adjudication, including whether the storage of copyrighted news content for training AI models amounts to infringement, whether AI-generated outputs infringe copyright, whether such use qualifies as fair dealing under Section 52 of the Copyright Act, and whether Indian courts possess territorial jurisdiction despite OpenAI’s servers and AI infrastructure being located outside India.
Considering that the dispute raised novel questions concerning generative artificial intelligence, copyright law and machine learning, and in the absence of any judicial precedent in India, the Court appointed Advocate Adarsh Ramanujan and Professor Dr Arul George Scaria of the National Law School of India University as amici curiae.
During the proceedings, Dr Scaria submitted that copyright law recognises a distinction between expressive and non-expressive use of copyrighted material and argued that AI training predominantly constitutes non-expressive use, which ordinarily does not amount to copyright infringement, while certain expressive uses may also be protected under statutory exceptions. Following extensive arguments from all parties over several months, the Court had reserved its interim order on March 27.
The post Delhi High Court holds OpenAI’s use of ANI content for ChatGPT training prima facie protected under fair dealing appeared first on India Legal.