Feb 27, 2024 - Technology

OpenAI says NYT "hacked" ChatGPT to produce allegedly infringing results

Illustration of a gavel on a sounding board with the copyright symbol on it

Illustration: Sarah Grillo/Axios

OpenAI on Monday asked a federal court to dismiss a lawsuit by the New York Times, maintaining the newspaper had to "hack" ChatGPT in order to get the system to produce allegedly infringing results.

Why it matters: Generative AI systems face a number of legal challenges that charge they infringe on copyrighted material both in the results they provide and in how they were trained. The results of these lawsuits could dramatically alter the future of generative AI.

The latest: OpenAI says in the new court filing that the New York Times had to exploit a bug, try tens of thousands of times and upload specific articles in order to get OpenAI's systems to return the verbatim passages that the newspaper cites in its complaint.

  • "The allegations in the Times's complaint do not meet its famously rigorous journalistic standards," OpenAI said in its filing. "The truth, which will come out in the course of this case, is that the Times paid someone to hack OpenAI's products."

OpenAI says that it has pledged to fix the bug in question, but notes that the Times still had to rely on deceptive prompts that violate its terms of use.

  • "Normal people do not use OpenAI's products in this way," OpenAI said in the filing.

Catch-up quick: The New York Times sued OpenAI and Microsoft in December, alleging that the companies trained their generative AI systems using articles without permission.

The big picture: The NYT lawsuit is one of a number of high profile cases against OpenAI, Microsoft and other AI companies. OpenAI also faces a lawsuit from a number of authors, while Getty has sued Stable Diffusion.

Yes, but: "The decision by OpenAI and other generative AI developers to enter into deals with news publishers only confirms that they know their unauthorized use of copyrighted work is far from 'fair," Ian Crosby, Susman Godfrey partner and lead counsel for The New York Times told Axios in a statement.

  • "What OpenAI bizarrely mischaracterizes as 'hacking' is simply using OpenAI's products to look for evidence that they stole and reproduced The Times's copyrighted works," Crosby wrote.

Editor's note: This story has been updated with comments from The New York Times.

Go deeper