- The Delhi High Court declined interim relief for ANI in its copyright case against OpenAI.
- The Court stated that OpenAI's use of ANI's works is "fundamentally different" from news articles and that ChatGPT responses are not substitutes.
- The Court observed that data scraped to train LLMs is used privately and not accessible to the public.
- This ruling is the first substantive judicial finding in India on whether AI companies can use copyrighted news content without a licence; similar lawsuits are pending in the United States and Canada.
The Delhi High Court's ruling on Friday marks a pivotal moment in India's first major AI copyright dispute, determining that OpenAI's use of ANI's content for training ChatGPT does not constitute copyright infringement.
Justice Amit Bansal stated that the use of copyrighted material falls under the fair dealing exemption for research as outlined in the Copyright Act. The court emphasized that the outputs generated by ChatGPT were not substantially similar to ANI's original articles, thus ruling out any infringement under Section 51.
*"ANI has failed to satisfy this Court that any memorisation or regurgitation of ANI's copyrighted literary works has happened to the responses generated by ChatGPT,"* the court noted, highlighting the lack of prima facie evidence for ANI's claims.
The court also pointed out that granting an interim injunction could hinder the growth of AI technologies in India, adversely affecting millions of ChatGPT users. *"Any interim injunction granted at this stage would, in my opinion, be detrimental to the growth of AI and more particularly, to the LLMs being developed in India,"* the order stated.
ANI, which was the first Indian media house to sue OpenAI, alleged unauthorized use of its content in a lawsuit filed in November 2024. The court's decision is significant as it sets a precedent for similar copyright lawsuits against AI companies, which are also pending in the United States and Canada.
The ruling underscores the ongoing debate over the use of copyrighted material in AI training and its implications for the media industry.
“The court observed that data scraped to train LLMs is used privately and not accessible to the public, distinguishing it from direct reproduction. This ruling marks the first substantive judicial finding in India on AI companies' use of copyrighted news, with similar lawsuits pending in the US and Canada.”
