Exclusive: OpenAI tells Indian court that removal of ChatGPT data violates US legal obligations, Reuters

By Arpan Chaturvedi, Aditya Kalra and Munsif Vengatil

NEW DELHI (Reuters) – OpenAI said in a recent filing to Reuters that any order to an Indian court to remove training data that powers its chat GPT service would be inconsistent with its legal obligations in the United States.

The Microsoft-backed AI firm also said it was not within the jurisdiction of Indian courts to hear the copyright infringement case filed by local news agency ANI as OpenAI has no presence in the country.

In the most high-profile and closely watched case against the use of AI in India, ANI sued Delhi-based OpenAI in November, accusing the news agency of using its published content to train ChatGPT without permission.

In an 86-page filing in the Delhi High Court on January 10, which was previously unreported, OpenAI responded to the suit seeking deletion of ANI data stored by ChatGPT.

OpenAI and other companies have faced similar lawsuits from prominent copyright holders for misusing their work to train AI models. New York Times (NYSE: ) on OpenAI in the United States.

OpenAI has repeatedly denied the allegations, saying its AI systems use publicly available data fairly.

In a hearing in November, OpenAI told a Delhi court that it would not use ANI’s content, but the news agency argued that the published works were stored in ChatGPT’s memory and should be deleted.

In a January 10 filing, OpenAI said it is currently awaiting litigation in the United States over the data on which its models were trained, where laws require it to retain the data pending hearings.

OpenAI is “therefore under a legal obligation under United States law to protect and not to delete said training data,” OpenAI said.

OpenAI did not respond to a request for comment.

In its statement, it said the relief sought by ANI was not subject to the process of Indian courts and was beyond the reach of the court.

The company “has no office or permanent establishment in India … the servers on which (ChatGPT) stores its training data are also located outside India”.

ANI, which has a 26% interest in Reuters, said in a statement that it believes the Delhi court has the jurisdiction to decide on the matter and will respond in detail.

A Reuters spokesman did not immediately respond to a request for comment, but said in November that the agency was not involved in ANI’s business practices or operations.

A New Delhi court will hear the case on January 28.

After raising $6.6 billion last year, OpenAI was preparing to transition from a nonprofit to a for-profit business as it sought additional funding to stay in the expensive AI race.

In recent months, he has signed deals with Time magazine, the Financial Times, Business Insider-owner Axel Springer, France’s Le Monde and Spain. hurry up (BME:) Media for displaying content.

ANI also expressed concerns about unfair competition given OpenAI’s business partnerships with other news organizations and told the court that in response to the user’s request, ChatGPT had done verbal or similar work to ANI.

In its rebuttal to OpenAI, ANI argued that ANI “attempted to use its own articles as a query to manipulate ChatGPT.”