close_game
close_game

‘Dismiss application of book publishers’: OpenAI to Delhi HC

Jan 27, 2025 04:57 PM IST

OpenAI argued that the examples cited by FIP do not show that OpenAI’s LLMs were trained on the “original literary works” of the publishers.

Tech giant OpenAI has urged the Delhi high court to dismiss the impleadment application filed by an industry body of book publishers in ANI’s copyright lawsuit against the platform.

Representational image. (Shutterstock)
Representational image. (Shutterstock)

The US-based artificial intelligence organisation categorically denied that the copyrighted “original literary works” of book publishers were used to train large language models (LLMs) underlying ChatGPT, the key argument that the Federation of Indian Publishers (FIP) had made in its impleadment application earlier this month.

OpenAI argued that the examples cited by FIP do not show that OpenAI’s LLMs were trained on the “original literary works” of the publishers. It denied that its web crawlers had sourced content from publishers that was not already publicly available online.

In the suit, FIP cited four examples to prove that ChatGPT had been trained on/had access to the members’ published, copyrighted books. For instance, on prompting ChatGPT, the chatbot gave summaries of articles in a peer-reviewed journal called ‘Journal of Psychosocial Research’ published by Prints Publications, and a detailed summary of ‘Financial Management’, written by Prasanna Chandra and published by PHI Learning.

In its submission to the court, OpenAI claimed that in the examples cited by FIP, the company’s chatbot ChatGPT had shown the publicly available online sources from which it drew the summaries. It also said that content such as abstracts, summaries and table of content were already publicly available on the publishers’ website. In a table, OpenAI showed that it had taken the information from either the official website of publishers, Amazon India, or websites such as Wikipedia and ResearchGate.

An OpenAI spokesperson, in a statement to HT, said, “We build our AI models using publicly available data, in a manner protected by fair use and related principles, and supported by long-standing and widely accepted legal precedents.”

"We take great care in our products and design process to support news organisations. We are actively engaged in constructive partnerships and conversations with many news organisations around the world, including India, to explore opportunities, listen to feedback, and work collaboratively. Along with our news partners, we see immense potential for AI tools like ChatGPT, including through search, to deepen publishers’ relationships with readers and enhance the news experience,” the spokesperson said.

Earlier on Monday, HT Digital Streams (digital arm of Hindustan Times), IE Online Media Services Private Limited (Express Group), NDTV Convergence, and industry body Digital News Publishers Association (DNPA) filed an intervention application in the Delhi HC in the same lawsuit against OpenAI, , arguing that their content had been scraped and used without permission or licence.

In its submission to the lawsuit filed by ANI in November, OpenAI argued that its web crawlers can only access publicly available data and are designed to not circumvent technical access controls such as logins or paywalls. It reiterated this argument in its response to FIP’s application. The company said it uses “only the non-expressive, statistical characteristics of data to train its LLM, and not the expression contained therein”. The latter is protected by copyright.

OpenAI said that FIP should not be impleaded as the industry body itself does not hold a copyright, and thus, the outcome of ANI’s lawsuit will not affect it. It also said that book publishers and news agencies such as ANI do not have common ground with respect to the questions of law raised in the ANI lawsuit.

A single bench of Justice Amit Bansal, in the November 18 order, had cited four questions: first, if storage of ANI’s news data to train OpenAI’s software was copyright infringement; second, if using this data to generate responses was copyright infringement; third, if OpenAI’s use of ANI’s data is “fair use”; and fourth, if Indian courts have jurisdiction when OpenAI’s servers are in the USA. FIP, in its application, had said that it could assist the court in tackling these four questions.

OpenAI also argued that the FIP — whose members include Bloomsbury India, Penguin Random House, Pan Macmillan India, Rupa Publications, Wiley India, and scores of Indian publishers of school and college textbooks, such as APC and S Chand — has not shown that it has been authorised to move this application by its members.

rec-icon Recommended Topics
Share this article
Get Current Updates on India News, Weather Today, Latest News at Hindustan Times.
See More
Get Current Updates on India News, Weather Today, Latest News at Hindustan Times.
SHARE THIS ARTICLE ON
SHARE
Story Saved
Live Score
Saved Articles
Following
My Reads
Sign out
New Delhi 0C
Thursday, February 13, 2025
Start 14 Days Free Trial Subscribe Now
Follow Us On