OpenAI contends New York Times hacked ChatGPT to file copyright lawsuit

New York Times hacked ChatGPT: OpenAI has requested a federal judge to dismiss portions of the New York Times' copyright lawsuit, contending that the newspaper engaged in what it terms as "hacking" of its chatbot, ChatGPT, and other artificial intelligence systems to produce misleading evidence for the case. 

In a filing submitted to Manhattan federal court on Monday, OpenAI asserted that the Times induced its technology to replicate its content via "deceptive prompts" that violate OpenAI's terms of use. OpenAI further alleged that the Times hired someone to manipulate its systems, though it did not specify the individual and did not accuse the newspaper of violating anti-hacking laws.

Reacting to OpenAI's claims, Ian Crosby, the attorney representing the Times, stated on Tuesday that what OpenAI characterises as 'hacking' is only using OpenAI's products to search for evidence of copyright infringement. Representatives for OpenAI have not yet provided a comment regarding the filing.

The legal dispute originated in December when the Times sued both OpenAI and its major financial supporter, Microsoft, alleging the unauthorised use of millions of its articles to train chatbots for information dissemination.

This lawsuit is part of a broader trend where copyright holders, including authors, artists, and music publishers, have taken legal action against tech firms over the alleged misuse of their content in AI training. Tech companies, however, argue that their AI systems fairly utilise copyrighted material, and they warn that such lawsuits could stifle the growth of the burgeoning multitrillion-dollar AI industry.

Notably, courts have yet to determine whether AI training constitutes fair use under copyright law. Some infringement claims related to AI-generated content have been dismissed due to a lack of evidence demonstrating similarity to copyrighted works.

The Times' complaint highlighted instances where OpenAI and Microsoft chatbots provided users with near-verbatim excerpts of its articles upon request, accusing the companies of attempting to capitalise on the newspaper's journalistic investment and create a substitute for its content.

OpenAI countered in its filing, stating that it took the Times "tens of thousands of attempts" to generate the alleged anomalous results, highlighting that ChatGPT cannot typically serve up Times articles at will.

Furthermore, OpenAI expressed confidence that it and other AI firms would prevail in their legal battles based on the fair-use question, asserting that preventing AI models from acquiring factual knowledge is akin to trying to restrict news organisations from re-reporting stories they did not investigate themselves.

(With Reuters inputs)