Judge Alsup Just Lit a Match Under the AI Licensing Industry.
Training = Fair Use. Cope Accordingly.
If you’re still waiting for the New York Times to blow up OpenAI and send Anthropic crawling back to your inbox begging to license your content — I regret to inform you:
The window has closed. The leverage is gone. The fair use train has left the station, and Judge Alsup is driving.
Read the Judge’s order here: analaw.com/anthropic-fair-use-order
Let’s break it down.
The Ruling, in One Sentence:
“Training LLMs on copyrighted books is exceedingly transformative, and therefore fair use.”
That’s it. That’s the tweet. That one sentence just detonated half of Legal Twitter’s doomer takes and every “maybe we’ll sue too” strategy memo still gathering dust in a media company’s Google Drive.
⸻
Quick Context (for the People Who Still Think This is a Vibe Shift and Not a Legal One)
Two days ago, Judge William Alsup — the same guy who learned Java during Oracle v. Google just to write a more brutal opinion — issued the first clear ruling on whether it’s legal to train a language model on copyrighted works.
His conclusion?
Learning is not theft. Reading is not infringement. Training is not copying.
He said it better. But the subtext was basically:
“I’m not torching generative AI so Random House can shake down OpenAI for using page 47 of some book no one read.”
⸻
If You’re a Content Owner Waiting to Cash Out on a Model’s Existential Crisis… You Missed It
Let me translate this into actual strategy:
The big AI labs were licensing content not because they had to, but because discovery was nasty and the outcome was uncertain.
Alsup just erased that uncertainty.
The strategic math flipped overnight.
Before:
“Buy a license, avoid a lawsuit.”
Now:
“Why pay for something a federal court just said we never needed?”
If you’re still holding out for a payday — spoiler:
You’re not negotiating anymore. You’re auditioning for free.
⸻
But Wait — What About the Pirated Books?
Yes. Anthropic apparently acquired some training data from Books3 and LibGen.
And yes, Alsup dragged them for that. As he should.
“Fair use isn’t a get-out-of-jail-free card for piracy.”
Cool. That matters — if you’re still training on bootleg PDFs in a basement somewhere. Most labs aren’t. They’re buying books, scanning them, documenting chain of custody like they’re prepping for a Hague tribunal.
So while the piracy point makes for good headlines, it’s legally orthogonal to the actual stakes:
Using copyrighted books — lawfully obtained — to train a model is now court-blessed.
If you’re confusing acquisition method with use case, you’re missing the ruling. Or intentionally misreading it, which is worse.
⸻
Outputs Still Matter. But Inputs Just Got Downgraded to B-roll.
This opinion doesn’t kill every copyright case. It just kills the ones hoping to win on training alone.
If you’ve got a model regurgitating entire chapters of your book verbatim?
Congratulations, that’s still in play.
If your theory was “they read my stuff and that feels icky”?
Pack it up. That’s not a cause of action. That’s a LinkedIn post.
⸻
Final Thought: The Leverage Window Didn’t Close. It Slammed.
This is the ruling. This is the precedent. This is the federal judge standing up and saying:
“We’re not dismantling an entire technological paradigm because someone trained a model on a book they legally bought.”
And he’s right.
Because if reading something and learning from it is illegal, I’ve got bad news for literally every student, lawyer, and software engineer on earth.
So no — you’re not getting a licensing windfall.
And no — the courts aren’t going to save you from the fact that generative AI learns like a human and outputs like a remix engine.
The only question left is whether you’re still pretending your corpus is leverage — or whether you’re ready to pivot.
My advice?
Pivot. Fast. Before the next judge turns this ruling into boilerplate.
⸻
🧠 If you’re still trying to figure out what’s ownable, enforceable, or just emotionally resonant but legally worthless — I do that work.
📬 DM me before your “strategy” turns into discovery.
⸻
🤖 Subscribe to AnaGPT
3x a week [MWF], I break down the latest legal in AI, tech, and law—minus the jargon. Whether you’re a founder, creator, or lawyer, my newsletter will help you stay two steps ahead of the competition.
➡️ Forward this post to someone working in/on AI. They’ll thank you later.
➡️ Follow Ana on Instagram @anajuneja
➡️ Add Ana on LinkedIn @anajuneja