OpenAI claims it does not 'steal' content from Indian media platforms to train ChatGPT: Report

AI giant OpenAI has denied accusations that it “steals” articles from major Indian media outlets and uses them to train ChatGPT, according to a Reuters report.

The report said that the Sam Altman-led company argued that its use of publicly available content does not require special licensing arrangements, even as local media giants — including Adani’s NDTV, Mukesh Ambani’s Network18, the Indian Express, the Hindustan Times, and the Digital News Publishers Association (DNPA) — attempt to join a lawsuit alleging that ChatGPT exploits their intellectual property without permission or payment.

The dispute dates back to last year, when Indian news agency ANI sued OpenAI, claiming that the U.S. firm was repurposing ANI’s articles to train ChatGPT without consent or compensation. Since then, other publishers and media groups have joined forces, each contending that the artificial intelligence model has “scraped” content from their websites.

The report further said that OpenAI’s recent filing, submitted on Feb. 11, seeks to prevent these parties from formally entering the legal battle. The company denies it “has used any of the applicants’ or the DNPA’s members content” to develop its chatbot’s capabilities.

OpenAI argues it relies on materials that are already posted online, pointing to widely recognized “fair use” and legal precedents allowing for such data collection.

Although the ChatGPT maker has inked licensing deals with some news outlets abroad, like The New York Times, the Indian publishers argue that none of them has received a similar offer. According to their legal submissions, OpenAI’s stance means Indian organizations are left uncompensated for any training data derived from their sites.

It has previously released statements indicating that it considers publicly posted information on the internet to be valid training data, consistent with existing laws and industry standards.