Lionel Nation - 🚨 JUDGE FORCES OpenAI to Hand Over 20 MILLION ChatGPT Chats in NYT Copyright WAR

All Episodes

Previous Next

Dec. 4, 2025 - Lionel Nation

25:42

🚨 JUDGE FORCES OpenAI to Hand Over 20 MILLION ChatGPT Chats in NYT Copyright WAR

Large

Audio Only |

Scroll

Time	Text
	Sometimes a story will pop up that seems not exactly the most shocking, but it is absolutely shocking.
	Sometimes there's a story that deals with ChatGPT, OpenAI, artificial intelligence.
	And per usual, a lot of people don't really feel the that you should, especially when it's cataclysmic.
	But a Manhattan magistrate just told OpenAI to hand over 20 million anonymized chat GPT conversations to the Daily News and a coalition of major news outlets, newspapers, that are suing the company for what they describe as industrial scale copyright infringement.
	This is huge.
	20 million chats, not 20,000, not a few curated samples, scrubbed by lawyers.
	20 million.
	And according to the court, that massive, immune number represents only 0.05% of the total logs that OpenAI has kept behind the curtain.
	Now, the hardest part, of course, is trying to get people to grasp the notion of this.
	It's going to be very, very difficult because it sounds a little dry, but it is not.
	This is not some casual discovery order.
	It is a blunt, brutal statement from the judiciary, from the court, from, remember, Lionel's law, the law always lags behind technology, but it is a blunt, brutal statement from the court that OpenAI's data practices and its reliance on copyrighted content are fair game,
	especially if the plaintiffs can show what they think this sample will show, that ChatGPT has been spitting out partial or complete reproductions of news articles and columns and other journalistic work.
	In other words, the outlets want to prove that the model is not learning abstract patterns by ingesting and digesting and regurgitating copyrighted pieces, that there's no learning.
	There's not this learning, but that they're basically collating all this stuff.
	And they're just kind of refabricating and reformulating stuff that's copyrighted from working reporters without permission or payment.
	This is huge.
	I don't know if we can say this.
	And the judge basically said, go ahead, dig in.
	And Magistrate Judge Ona Wang, or sometimes it's pronounced Fong, I'm not sure, but in any event, Judge Ona Wang did not buy OpenAI's plea to rethink her earlier ruling, not even close.
	She upheld her order and emphasized that the data are relevant, proportional, and necessary.
	The privacy argument, which OpenAI has been, you know, leaning on like a crutch, didn't impress her in the least.
	She made it very clear that OpenAI itself admitted it is nearly done anonymizing the logs.
	She also underscored that the case already has multiple privacy protections in place because of the sensitive nature of the exchanged material, meaning that OpenAI cannot hide behind user privacy when what they really fear is what the logs will expose about how the, dare I say that terrible expression, how the sausage gets made.
	Now, listen to what I'm saying, my friend.
	This is again incredible.
	This is a, oh no, it's here.
	And understand what they've been telling you is that, oh no, we're not stealing this stuff.
	We're just learning this.
	Yeah, that's it.
	We're learning this.
	We're not just spewing and regurgitating your stuff.
	We're learning this.
	Well, let me stop for one second, my friend, because there's something which is very, very critical.
	And I can't say this enough.
	And I think you know me enough.
	I have been the biggest believer in food and emergency food preparedness since day one.
	And very, very important for you right now, especially around the Christmas season.
	This is no time, no time for, hey, everything's great, for complacency.
	Nope, not at all.
	And that's something which I want you to understand.
	There is something very, very critical here.
	Very, very critical.
	And that seemingly is this.
	Simply is this.
	Unless you do something or recognize what you have to do, it could be very, very too, it could be too late.
	Right now, there is something called, it's wonderful.
	It's called give the gift that actually matters this year.
	It's called buy one.
	Okay.
	Listen, buy one.
	I forgot my, where's my line here?
	Oh, yes.
	I had a brain for it.
	Buy one gift to.
	Buy one gift to.
	This has been a long day for me.
	Buy one gift to.
	This is a Christmas special.
	And you need to hear this very, very clearly.
	All month long, when you pick up a four-week emergency food kit, they are going to send you two additional one-week food kits, absolutely free.
	No strings, no gimmicks, no surprises.
	That's it.
	You secure your household's backup supply and you instantly have two life-saving gifts to hand to people that you care about and you love.
	That's it.
	Simple.
	And it sure beats a sweater.
	Those one-week kits are perfect for anybody already in preparedness who knows what they're doing.
	Or somebody, maybe you want to nudge along.
	You want to kind of push them and you want to light the fire.
	You want to give them a wake-up call.
	One small box can turn someone from vulnerable to ready.
	And in times like these, as you know, now's the time to do it.
	And if you think we're out of the woods right now, you're out of your mind.
	One mistake, one supply chain, supply chain screw up and we're done.
	So anyway, we all feel the tension in the air right now.
	We all see how quickly shelves empty when anything unexpected hits.
	And especially as this nor'easter hits and people right now are up to their arson snow, having food put away is no longer fringe.
	It is responsible and it is sane.
	And this buy one gift to deal means you're protecting yourself and your family while actually giving something meaningful for once for Christmas.
	That simple.
	So go to preparewithlionel.com, preparewithlinel.com.
	That simple, preparewithlinel.com and go there and do what I've been telling you for the longest time.
	It makes complete and total and absolute sense.
	You dig?
	Okay.
	Now, I was doing an interview today with a very nice young man.
	I'll announce it later when it comes out.
	And I said the number one, the number one existential threat that poses here this world, this society is AI.
	And it's something which nobody really understands.
	And the hardest part is that no matter how many times I try to say, let me try it this way.
	Let me try to explain it this way.
	People don't really understand what it means.
	But for the first time now, people are fighting back.
	And this is huge.
	See, Google, whenever it collated something, whenever it did something, this is very, very important.
	If Google or Yahoo or somebody took articles, they would say, here's the article, but it's the person's name and copyright and basically actually helped in disseminating information.
	It didn't rewrite it.
	It didn't rewrite it and put it together as its own.
	And that's kind of what we're talking about right now.
	It's something which is very, very scary.
	So for the first time, for the first time now, all of a sudden, people are going to be asking the question: what exactly does AI do?
	What does AI do?
	That's the thing which is most impressive.
	Did it really learn, or are you just stealing and refabricating and refurbishing stuff which has been heretofore copyright protected?
	Let me tell you something.
	The sense of the fear in San Francisco is huge.
	I mean, not dread exactly, but a deep, the deep focus concerns like, uh-oh.
	Because once those logs go out, even anonymized logs, the plaintiffs will get a microscope into OpenAI's secret garden.
	What do they do?
	This trade secrets?
	What's in there?
	And finally, someone outside the company will get to peek into how often this model spits out copyrighted texts and how closely, and whether, by the way, the system uses that content to train itself in ways that the company has long danced around in public statements.
	In fact, you're going to be seeing other AI companies watching very, very carefully because this is a game changer.
	See, this is the same way.
	This is the same way that we're going to be asking the question, which I've been asking for a long time.
	If AI is able to produce, for example, something that will normally be considered CSAM, child sexual abuse material, and if it were to create it in such a way that it looked absolutely exactly like the real thing, so to speak, would it be subject to prosecution?
	Is it the memorialization of a crime scene, or instead, is it something that's different?
	Is it something that looks like something that's illegal?
	Now, that may not sound like much to you, but it's enormous.
	What it means is it changes everything.
	Because remember what I'm telling you, the law always lags behind technology.
	And why is this a huge moment?
	Let me tell you.
	It's a huge moment in the consolidated class action lawsuit that began in 2023 because it's not just the New York Times anymore.
	Now you have the Daily News, you got Tribune Publishing, Media News Group, and a lineup of authors who have gone from curious to furious, as they say, about how big AI uses their work.
	Nobody really knew.
	And they argue that OpenAI in particular built an empire using content it never owned.
	And now it faces, well, scrutiny and inspection like you can't believe.
	Open AI has been accused by the judge and the plaintiffs of dragging his feet, withholding key evidence, and also attempting to stall this thing.
	And there's this fellow, Frank Pine, executive.
	Notice how he said that, Pine.
	Frank Pine, executive editor for Media News Group and Tribune Publishing.
	He said that OpenAI was hallucinating if they ever thought they could get away with withholding evidence about a business model that steals from hardworking journalists.
	I mean, that's not subtle.
	That's a shot across the bow.
	And what is this going to do for AI music?
	See, you can hear the message between the lines.
	They think OpenAI is nervous because the logs might show wholesale copying.
	They don't know what it's going to show.
	And there's no way you can hide or destroy the evidence because that's called spoliation.
	You see, in the Wednesday order, Wang, Judge Wang Fong, called the conversations clearly relevant for two reasons.
	First, because they may indeed contain copied material.
	But second, because OpenAI claims that they didn't, or they do not rather, do that, and that user input explains any suspicious output.
	So the only way to test both claims is to see the data.
	That's basic civil procedure.
	If you are accused of misusing copyrighted material, if that's a story, if discovery is presented, well, discovery then is the only way to focus on the guts of the operation.
	And they're going to also claim trade secrets.
	But here's where it gets even more interesting.
	The judge pointed out that OpenAI has tens of billions of logs, tens of billions.
	And those are logs retained in the normal course of business.
	So maybe the company keeps user data on a scale few people fully appreciate.
	Suddenly, suddenly those privacy claims look a little less urgent.
	See, if OpenAI can store tens of billions of user outputs without legal worry, then handing over 20 million anonymized slice versions, giving them to plaintiffs under a protective order is hardly a threat to anyone.
	At least that's the argument.
	And the decision, by the way, also comes with a quiet, but a very strong reprimand.
	In her order, Judge Wang noted that OpenAI had withheld critically important evidence earlier in the case.
	And that stings.
	You don't want the discovery violations, withholding information, this is critical.
	See, courts don't toss language like that around lightly.
	And she went even further.
	She raised the question that OpenAI's delays were motivated by an improper purpose.
	Uh-oh.
	Then she added, then she said that neither possible explanation for their conduct look good.
	See, that is a judge speaking politely but clearly.
	So she's telling OpenAI she knows when a litigant is playing games and they're playing games.
	Now, the attorney for the news organization, Stephen Lieberman, amplified that point by saying that the ruling essentially confirms that OpenAI attempted to dodge transparency, no other way around it.
	The plaintiffs now feel they have the upper hand.
	And in today's political climate, where trust in big tech is non-existent and where populists on their right have spent years warning about unregulated AI power grabs, this is shaping up as a major, colossal flashpoint.
	This is bigger than, oh my God.
	Because they did this and they said, get it out, and then we'll worry about this stuff later.
	But here's the irony.
	This is an interesting part.
	Open AI insists that their system doesn't store or reproduce training data.
	Yet, yet, the very existence of billions of user logs creates the opposite impression.
	And the company has already acknowledged that the model itself can sometimes output copyrighted passages.
	So, if these logs show patterns of repeat copying, it could then open the door to a tsunami of legal arguments that these models were built in violation of copyright rules that were not written with AI in mind.
	That's the battlefield.
	Old law meets new machine.
	Huge.
	So, Silicon Valley has long operated with a kind of a kind of a frontier mentality.
	Innovate first, ask permission later.
	Hope the courts never catch up.
	It's almost like, think about this.
	The very camera that you have has refocused and repackaged how people deal with the notion of privacy.
	How many times have you heard people say, can you take a picture of me?
	Can you take a picture of me in public?
	Is this open view or do you can you do I have to can you do that?
	We never had this before.
	We had cameras.
	Remember when what was the name of that?
	What was the name of that file sharing?
	Remember all the music?
	What was the name of it?
	What was the name of it?
	When people would copy, you know the name of it.
	When people would send music back and forth, they would all of a sudden you get free albums and free songs.
	It was legit.
	They shut them down and then Spotify came along.
	They did a legitimate Napster.
	Thank you.
	Thank you, poy dog.
	Poi dog, by the way, is that poy dog pondering, the group?
	Spending the day in the shirt you wore.
	Is it interesting, interesting.
	Anyway, so this attitude then works until one judge says, and that's what happened.
	So this ruling is not just some procedural decision.
	It's a moment when the judiciary, when Article III courts place a very serious stake in the ground and reminding tech giants that the law still outranks their, you know, their venture capital little swagger.
	Now, the plaintiffs, the news folks, for the most part, believe that this is a watershed moment.
	If they can prove that ChatGPT contains direct lifts from news articles, then they have a strong case that OpenAI built its flagship product on other people's intellectual property.
	And then, who knows?
	How do you assess damages of that?
	And what does it do to the whole process?
	And if they can prove that OpenAI tried to hide that fact, things will get even uglier.
	It's one thing that you were liable, but you tried to disguise it.
	You tried to defraud the court.
	This is not about nostalgia for the old newsroom.
	It's about whether a multi-billion dollar AI company used protected work as free, like fuel, you know, fodder, lumber, so to speak, to build the structure of this.
	You want to talk about exploitation?
	Well, here it is.
	You want to talk about disruption?
	This is it.
	And if you want to talk about accountability, well, the good Judge Wang just dropped the hammer.
	So the logs are going to be handed over once the de-identification process finishes.
	Then the plaintiffs get seven days of waiting before they start gigging, and they'll probably use Chat GPT or AI to go through and sift through it.
	And when they do, the entire industry is going to hold its breath because those logs, those logs turned over, will become the most scrutinized data set in the AI world.
	And other companies are going to know what to show and what not to show.
	And Silicon Valley, by the way, built the future.
	Well, now the courts got to decide whether they did it legally.
	So let me just tell you something.
	The biggest problem I have sometimes is trying to explain to folks kind of what is important.
	As I have said, I don't know what my job is.
	I don't want to be an entertainer and just throw out, hey, this is a story that'll get clicks, which I guess we all should.
	But I want to find stories and issues that are critical, that inspire anger, that inspire absolute shock.
	And sometimes, especially when it comes to matters of AI, I don't get that reaction.
	I don't, it's very difficult to get people to understand and very difficult for people to feel this compelled focus.
	You see, I had this idea years ago when I first, I thought, when talk radio was in its nascent period, sort of, at least for me, that I felt like I was standing on the beach and I would say, come on, boys.
	And then the, you know, the throngs of humanity would surge, you know, because I would inspire people.
	I would, I would say something, come on, let's go.
	And then, you know, no, I don't get that.
	I just did a live recently on George Soros and nothing.
	It's almost like we really don't want to do anything about it.
	We would just, we kind of like to talk about it.
	You talk about it, but we don't really want to do anything.
	We don't really, we're just going to, meanwhile, the radical left, oh, they're, they stop at nothing.
	See, that's the thing where you have to think more like them.
	And whenever you have something that's really arcane and recondite, like AI, you really get to be problematic because most people don't understand what it is.
	You really still don't want to get it.
	Look, even with the atom bomb, people kind of knew they wouldn't know, you know, fission versus fusion versus the hole in the ground, but they knew that a bomb was the operative and critical word there.
	They understood that even during even during COVID, they didn't understand the means of transmissibility, but they understood the notion of viruses.
	They understood about, you know, virions and spike proteins.
	And they kind of had an idea, sort of.
	So people are definitely not, they're not stupid.
	But if it does not hit that level of fascination, look at this, this wonderful Bextrum, I believe her name is.
	See, even I forgot the National Guard sales up soldier, nothing.
	Epstein, nothing.
	Trump's assassination attempts, nothing.
	Nothing.
	Nothing ever done.
	Nothing follow through.
	No follow-up.
	No follow-up.
	And that is very, very disconcerting.
	But I'm not going to stop, my friends.
	I'm not going to stop.
	I'm going to still tell you what I think is important.
	And I'm going to tell you what I think is critical.
	I appreciate the fact that you're listening to me and that you're on the show and you're on our channel.
	And I want you to understand that this is like you being a part of the first, not being a part of the first teletype or telegraph or post office.
	No, no, no.
	It's like being a part of the invention of the I.
	I mean, this is not ChatGPT.
	This is a structure that on its own is going to take off.
	And by the way, a lot of this is still hand-fed to the AI, the ghost and the machine kind of an idea.
	But when it goes AGI and it goes completely, absolutely independent in terms of its own power, then you're not going to be able to stop it.
	You're going to say, look, we can't turn this off.
	It is through recursive self-improvement.
	It just made itself resistant to our passwords.
	It just separated itself and became 10 times stronger, which is even more, people don't understand about a machine that makes itself stronger and then breaks away from the developer of the machine.
	This is impossible for people to grasp.
	It's counterintuitive.
	And anyway, but they'll find out.
	In any event, dear friends, thank you.
	Please like this video.
	Subscribe to the channel.
	Make sure you subscribe, subscribe, subscribe.
	Comment.
	I love your comments.
	I'm going to put some questions up for you to think of.
	Please follow Mrs. L at Lynn's Warriors.
	And also, overnights, WABC, 1 to 5 a.m., 770, 770, 770 a.m.
	You can go to wabcradio.com.
	You can get the app.
	You can venture forth and connect in a variety of different ways.
	77 WABC.
	Call us.
	Be a part of it.
	Tell them you're a part of the nation.
	And until then, my friend, I say to you, as I always do, thank you so much for watching.
	And please comment as you see fit.
	And as they also say, the monkey's dead and the show's over.

▲