New Delhi: The New York Times and The New York Daily News have escalated their two-year copyright battle with OpenAI, filing a motion in a Manhattan federal court asking the judge to sanction the AI company for allegedly withholding and destroying evidence during discovery.
What is the case, exactly?
The New York Times sued OpenAI and Microsoft in December 2023, arguing that OpenAI trained ChatGPT on the Times’ journalism without permission or payment, and that the chatbot could reproduce that journalism in its outputs. Several other news organisations filed similar suits, which were later consolidated into one case. The Daily News and seven sister papers are part of that consolidated group.
At the centre of the dispute is whether OpenAI’s use of copyrighted news content to train its models counts as fair use or whether it amounts to copyright infringement. To make that case, the publishers have spent nearly two years trying to get access to OpenAI’s training datasets and ChatGPT conversation logs, to check how much of their content shows up in the system and how often it gets reproduced.
What does the new filing claim?
According to the motion filed Thursday, the publishers allege OpenAI “chose obstruction” instead of handing over that evidence early in the case. Their specific claims include:
OpenAI allegedly deleted billions of ChatGPT outputs after the lawsuit was filed, which the publishers say violated the court’s order to preserve evidence. Reportedly, when OpenAI eventually submitted a sample of chat logs in December, it came with so many redactions that the court called it “unusable.”
The publishers originally asked for a sample of 120 million chat logs. OpenAI negotiated that down to 20 million, and allegedly substituted millions of logs within that smaller sample.
In an April deposition, OpenAI data privacy engineer Vinnie Monaco reportedly said the company had already run internal searches of its training data for copyrighted journalism, something OpenAI had previously said it could not do.
The same deposition reportedly revealed that OpenAI had built an internal database of about 78 million de-identified ChatGPT conversations, used to gauge how often it was infringing on other parties’ work, along with a tool called a “Bloom filter” under an internal project named “Project Giraffe” that tracked regurgitation in outputs.
According to Reuters, Daily News attorney Steven Lieberman said the motion is meant to “punish OpenAI for hiding and destroying evidence showing how ChatGPT was trained on stolen journalism.”
The same Reuters reports suggests that responding to the sanctions motion, OpenAI spokesperson Drew Pusateri said that as the Times’ case has weakened and claims against the company have been dropped, the publishers are now making “blatantly false allegations.”
According to media reports, the Times has reportedly spent more than $28 million so far on litigation against AI companies, including a separate suit against Perplexity. The case is being watched closely because its outcome could shape how other publishers and rights holders approach licensing negotiations.