ChatGPT data removal in India will breach US legal obligations, OpenAI tells Delhi High Court

OpenAI informed an Indian court that any directive to eliminate the training data supporting its ChatGPT service would conflict with its legal responsibilities in the United States, based on a recent document reviewed by Reuters.

The AI company supported by Microsoft stated that Indian courts lacked jurisdiction to address a copyright infringement case filed by local news agency ANI, as OpenAI had no operations within the nation. In the most significant and monitored lawsuit regarding AI usage in India, ANI took legal action against OpenAI in Delhi in November, alleging that it used the agency’s published material without consent to train ChatGPT.

OpenAI addressed the lawsuit, which additionally aims for the removal of ANI’s data that ChatGPT has already stored, in an 86-page document submitted to the Delhi High Court on January 10, a detail that has not been reported before.

OpenAI and various companies are encountering numerous lawsuits from notable copyright holders regarding the supposed improper use of their work to train AI models, including one filed by the New York Times against OpenAI in the U.S.

OpenAI has consistently rejected the accusations, stating that its AI systems utilize publicly accessible data fairly.

In a November hearing, OpenAI informed the Delhi court that it would cease using ANI’s content, but the news agency contended that its published materials were retained in ChatGPT’s memory and needed to be removed. In the January 10 filing, OpenAI mentioned that it is presently engaged in litigation in the United States regarding the data used to train its models, where regulations mandate the preservation of the data while hearings are ongoing.

OpenAI “is thus legally required, according to United States laws, to maintain and not remove the specified training data,” it stated. In its filing, OpenAI further stated that the relief sought by ANI was not within the scope of Indian court processes and was outside their jurisdiction.

The business possesses “no office or permanent establishment in India … the servers that (ChatGPT) uses to store its training data are also located outside of India.” ANI, in which Reuters possesses a 26 percent stake, stated that it believes the Delhi court has the authority to rule on the issue, and it plans to submit a comprehensive reply.

A spokesperson for Reuters did not promptly reply to a request for a comment, but the agency stated in November that it was not engaged in ANI’s business practices or activities.

The New Delhi court is scheduled to review the case on January 28. OpenAI is preparing to shift from a non-profit organization to a for-profit company as it aims to attract additional funding to maintain its lead in the expensive AI competition after securing $6.6 billion last year.

In the past few months, it has finalized agreements with Time magazine, the Financial Times, Axel Springer (owner of Business Insider), France’s Le Monde, and Spain’s Prisa Media to showcase content. ANI has expressed its worries about unfair competition due to OpenAI’s commercial agreements with other news outlets, and has informed the court that in reply to user requests, ChatGPT has reproduced verbatim or largely similar excerpts from ANI’s content. In its response, OpenAI claims that ANI “has attempted to utilize exact excerpts from its own article as a prompt, aiming to influence ChatGPT”