Lawyers representing the New York Daily News and an array of news organizations suing OpenAI for allegedly stealing and distorting their reporters’ work have asked a Manhattan judge to sanction ChatGPT’s parent company, alleging the tech behemoth deleted millions of conversations they were required to hand over as evidence of copyright infringement.

OpenAI continued to destroy output logs despite orders from two judges to preserve and provide them to the news organizations, new court filings allege. More than 1 million logs that had been requested — containing information the news outlets believe was based on their journalists’ reporting — were subbed out, according to court documents.

“[A]fter this Court ordered OpenAI to produce 20 million logs over OpenAI’s vociferous and repeated objections, OpenAI substituted millions of conversations that it was ordered to produce with other conversations – seemingly because it had deleted millions of the selected logs,” attorney Steve Lieberman wrote to the court in a Monday letter.

“OpenAI has refused to answer News Plaintiffs’ questions about the deleted and substituted logs.”

Related Articles

Actor Awards, formerly SAG Awards, will announce nominations Wednesday

New dietary guidelines urge Americans to avoid processed foods and added sugar

Job openings slide to 2nd lowest level in 5 years as hiring remains sluggish

Wall Street’s hot start to the year cools as stocks drift

Horror of school attack in Uvalde, Texas, brings tears as officer faces trial over police response

The dispute has come up amid a complex lawsuit brought by The New York Times, The New York Daily News and other outlets affiliated with Tribune Publishing and MediaNews Group. The news organizations allege OpenAI is stealing and distorting their copyrighted works, thus providing ChatGPT users stolen reporting that’s often inaccurate. They have been joined by the Authors Guild, and a litany of best-selling writers are also parties in the complex litigation.

As part of the litigation, OpenAI was required to hand over the logs per a November order by Manhattan Magistrate Judge Ona Wang, which was affirmed this week in a Jan. 5 order by Manhattan Federal Judge Sidney Stein.

The A.I. company also engaged in “hashing,” meaning they changed the ID numbers of some 20 million anonymized ChatGPT user conversations, presenting an enormous challenge for the media lawyers sorting through the immense volume of materials, attorneys for the news organizations said.

In denying OpenAI’s objections to Wang’s order, Stein on Monday wrote that her “rulings were neither clearly erroneous nor contrary to law.”

“She adequately balanced ChatGPT users’ privacy interests against the relevance of the documents in light of the privacy protections already in place,” the judge wrote.

In challenging Wang’s November order, OpenAI had argued it was “clearly erroneous” for the judge to have rejected the company’s proposal to run search terms across a sample of 20 million anonymous chats, claiming it would better protect users’ privacy.

Stein said those arguments were “largely a repackaging” of the company’s failed argument that Wang failed to account for the privacy of ChatGPT users.

Lawyers for the news outlets say OpenAI also included billions of “grossly overbroad and inappropriate” redactions in the info they handed over, blacking out the names of news outlets cited to ChatGPT, bylines and other information critical to the case.

“Since this case was about our requests for content of certain publications,” Lieberman said in an interview, “It makes it rather hard to find the evidence we’re looking for.”

The news organizations’ request to the court asks the court to order the tech company to explain why it shouldn’t be held in contempt. They’re requesting an evidentiary hearing in the coming months.

The Daily News has reached out for comment to OpenAI, which did not immediately respond.

On Friday, a company spokesperson told the legal news outlet MLex, “As we’ve made clear to the Court, this is just another example of the Times distorting the facts and misrepresenting how our technology actually works.”