Thursday brought another round. Not just noise, but actual accusations. Publishers claim OpenAI is hiding evidence about how they train their AI models. A new motion drops into existing copyright lawsuits, and it’s heavy on allegation, light on apology.
The Accusations
Seventeen publishers filed this motion. You know who’s there? The New York Times. The Chicago Tribune. New York Daily News. Even Ziff Davis, which owns CNET. Ziff Davis started their own fight in 2025. They say OpenAI scraped their work to feed ChatGPT. Millions of articles, just taken.
This isn’t new territory. The Times started the legal war back in 2023. They sued OpenAI and Microsoft together. Claim: built tech on stolen journalism. Both companies said no way.
But this latest move isolates OpenAI. Microsoft is left out of these specific sanctions. The court is being asked to punish the AI firm. Why? Alleged obstruction. Datasets. Output logs. Evidence that allegedly proves how the training happened, now conveniently missing. If the judge agrees? Financial penalties loom.
“This motion asks the court to punish Open AI for hiding and destroying evidence…”
That’s Steven Lieberman, attorney for Daily News. He called it “stolen journalism.” Strong words. But the core issue remains the same. Generative AI spits out content. Sometimes verbatim. Sometimes summaries so close they blur lines. Expressive style mimicry. It feels like plagiarism, even if the law is still catching up.
Bleeding Revenue
Why care now? Money. Well, traffic, mostly. Digital media is hemorrhaging readers. AI overviews sit at the top of search results now. Users click. They get an answer. They leave. No visit to the actual site. No ad view. No subscription bump.
Publishers are terrified. Small outlets? Hit hardest. Some report traffic drops of 60%. Scary. One prediction says overall declines will top 40% by 2s29. Chatbots aren’t just helpful. They’re siphoning audience. Loyalty dies when the interface changes.
Ziff Davis claims OpenAI monetized their content on a massive scale, without permission. Lance Koonce, their lawyer, alleges more than theft. He says OpenAI lied about searching its own datasets. Serious misconduct. Or so the complaint goes.
Fair Use? Privacy?
OpenAI pushes back hard. They cling to “fair use.” Standard defense for training models on public internet data. But when publishers sue? The tone shifts.
An OpenAI spokesperson denied everything. Called the allegations “blatantly false.” Claimed The Times was invading privacy of innocent users. Wait, privacy? In a copyright suit? It’s a strange pivot. Defend user data while accusing publishers of weak cases. They insist The Times dropped some claims, so the foundation is cracking.
Yet, in a 2024 response to The Times’ original suit, OpenAI did something interesting. Accused the paper of deleting their own data. Data showing internal OpenAI usage? A counter-accusation of obstruction. The Times did drop one claim, true. But the main lawsuit? Still going.
Will the data turn up? Will sanctions stick? Probably not neatly. Copyright law moves slow. Tech moves fast. Someone usually pays the price for that gap. 📉
