Exclusive | OpenAI tells Indian court ChatGPT data deletion would violate U.S. legal obligations

NEW DELHI (Reuters) – OpenAI has told an Indian court that any order to delete training data powering its ChatGPT service would be inconsistent with its legal obligations in the United States, according to a recent document seen by Reuters.

The Microsoft-backed artificial intelligence company also said it was outside the jurisdiction of Indian courts to hear a copyright infringement case brought by local news organization ANI because OpenAI has no operations in the country.

In the most high-profile and closely watched lawsuit over the use of artificial intelligence in India, ANI sued OpenAI in Delhi in November, accusing it of using content published by the news agency to train ChatGPT without permission.

OpenAI responded to the lawsuit in an 86-page filing with the Delhi High Court on January 10, which also seeks deletion of ANI data that ChatGPT has stored, which was not previously reported.

OpenAI and other companies have faced a wave of similar lawsuits from prominent copyright holders for allegedly misusing their works to train artificial intelligence models, including a lawsuit filed against OpenAI in the United States by The New York Times.

OpenAI has repeatedly denied the accusations, saying its artificial intelligence systems use publicly available data fairly.

At a hearing in November, OpenAI told a Delhi court that it would no longer use ANI’s content, but the news agency argued that its published work was stored in ChatGPT’s memory and should be deleted.

OpenAI said in a Jan. 10 filing that it is currently defending a U.S. lawsuit over its model training data, which U.S. law requires it to withhold during the hearing.

“Accordingly, OpenAI has a legal obligation under U.S. law to retain and not delete said training data,” the statement said.

OpenAI did not respond to a request for comment.

OpenAI also said in its submission that the relief claimed by ANI is not subject to Indian court proceedings and is beyond the jurisdiction of Indian courts.

The company “has no offices or permanent establishment in India…(ChatGPT’s) servers that store its training data are also located outside India.”

ANI, in which Reuters holds a 26% stake, said in a statement that it believed the Delhi court had the jurisdiction to rule on the matter and would submit a detailed response.

A Reuters spokesman did not immediately respond to a request for comment, but the agency said in November that it had no involvement in ANI’s business practices or operations.

A New Delhi court is scheduled to hear the case on January 28.

OpenAI has been preparing to transform from a nonprofit to a for-profit enterprise, and after raising $6.6 billion in funding last year, the company hopes to obtain more funds to stay ahead of the costly artificial intelligence race.

In recent months, it has signed deals with Time magazine, the Financial Times, Business Insider owner Axel Springer, France’s Le Monde and Spain’s Prisa Media to display content. protocol.

ANI also said it was concerned about unfair competition given OpenAI’s commercial partnerships with other news organizations, and told the court that following user prompts, ChatGPT copied verbatim content or substantially similar excerpts of ANI’s work.

OpenAI argued in its rebuttal that ANI “attempted to manipulate ChatGPT using verbatim excerpts of its own articles as prompts.”

(Reporting by Aditya Kalra; Editing by Kirsten Donovan)