The Copyright Office Got Purged.

But the Real Crisis Is the Law Itself.

In light of Trump’s AI Action Plan being released this week, we need to talk about what he did to the copyright office this year.

Back in May 2025, the U.S. Copyright Office released its third and final report in a series on AI and copyright. It focused on whether using copyrighted content to train AI models is legal.

The conclusions weren’t radical. But they were inconvenient.

And within days, the head of the Copyright Office was fired.

This wasn’t just a bureaucratic swap.

It was a targeted removal.

And the implications reach far beyond D.C.

Because here’s the real headline:

The Copyright Office got it wrong. Trump just made it worse.

⸻

THE OFFICE BUILT A FRAMEWORK NOBODY CAN USE

Let’s start with the substance. The Copyright Office released three reports in under a year:

1️⃣ Digital Replicas (July 2024) — a call for a federal right of publicity.

2️⃣ Copyrightability (Jan 2025) — guidance on what counts as human authorship.

3️⃣ Generative AI Training (May 2025) — analysis of whether training models on copyrighted works qualifies as fair use.

Each report tackled real questions. But the answers? Half-law, half-laminate. They’re cautious, abstract, and ultimately unusable in the places where they’re needed most: courts, contracts, startups, and lawsuits.

Most critically, the Copyrightability report didn’t engage with how AI is actually being used. It treats a single prompt like a magic wand—press button, get words—and assumes the human is a passive bystander. It says that even detailed prompting doesn’t constitute authorship if the AI “controls the expression.”

That sounds legally tidy. It’s also technologically absurd.

If you’ve ever written anything meaningful with AI, you know it’s not “one prompt, done.” It’s recursive. You edit. You regenerate. You tweak tone, structure, and language. You build style constraints. You may iterate 20 or 200 times. That process isn’t a roulette wheel—it’s authorship.

And yet, the Copyright Office drew a sharp line: if a model “contributes too much,” your copyright claim might evaporate. No clear test. No guidance on thresholds. Just a murky suggestion that using AI might forfeit your rights.

Which raises the very real risk: we’re about to enter a decade of litigation over whether otherwise creative, original, AI-assisted works are even protectable.

⸻

HERE’S WHAT THAT LOOKS LIKE IN PRACTICE

A founder writes a product guide with ChatGPT.

A novelist co-authors a book using Claude.

A marketing team uses Midjourney to generate campaign visuals, then layers text and layout manually.

They register the work. A competitor copies it. The infringer claims: “The original wasn’t copyrightable anyway—it was AI-assisted.”

Now what?

Do we audit prompt logs? Do we subpoena draft versions? Do we litigate over what counts as “meaningful human control”?

There is no usable standard. That’s the Copyright Office’s failure.

⸻

THEN TRUMP BLEW UP THE BUILDING

On May 10, 2025—less than 24 hours after the Copyright Office released its third AI report—Trump fired Shira Perlmutter, the Register of Copyrights. He also removed Carla Hayden, the Librarian of Congress who had appointed her.

No official reason was given. But the timing wasn’t subtle. The report had pushed back on Silicon Valley’s favorite argument: that training on copyrighted data is just “reading” and therefore fair use. The Office didn’t say training was illegal. But it didn’t bless it either.

And that was enough. Trump removed both officials and installed loyalists.

According to reports from CBS News, Politico, and legal observers, the Trump Administration bypassed the standard process and quietly appointed acting leadership drawn from DOJ and Library of Congress legal staff—none of whom were confirmed or announced through the usual public channels.

The new acting Register was reportedly a DOJ deputy with no prior experience running the Copyright Office, and the interim Librarian was a former White House legal advisor. Their names were not disclosed, and the appointments raised constitutional questions, since the Librarian of Congress is a legislative branch role that typically requires Senate confirmation.

The message was clear: If your legal interpretation creates friction for AI development, you’re gone.

Here’s the more shocking part: almost no one covered it.

No mainstream media stories. No network news pieces. No congressional hearings or op-eds on the institutional damage. In the middle of the most consequential copyright debate in decades, the firing of the nation’s top copyright official barely made a ripple.

That silence isn’t neutral. It’s dangerous.

Because when core legal institutions can be politically purged—without press, without protest, without scrutiny—that sets a precedent. One where copyright law isn’t shaped by courts or Congress, but by executive pressure and media indifference.

We’ve now lost not just guidance, but governance.

And no one seems to care.

⸻

TWO WRONGS, NO RIGHTS

Let’s be blunt.

The Copyright Office misunderstood how AI works. It mischaracterized prompting, downplayed iterative authorship, and failed to offer real legal standards for hybrid human–AI creativity. The reports were cautious, doctrinal, and largely disconnected from the reality of building or using generative tools.

But firing its leadership didn’t fix that. It made it worse.

Now we don’t just have unclear doctrine. We have doctrine under political siege.

It’s not just that the Office got AI authorship wrong.

It’s that now, anyone trying to fix it risks getting fired.

⸻

THIS ISN’T JUST A LEGAL DEBATE. IT’S A COMING WAR

Within five years, courts will be flooded with cases over AI-assisted works.

Did the user prompt too much?

Did the model contribute too heavily?

Is this asset protectable, or public domain by default?

Bad actors will weaponize uncertainty. They’ll knock off AI-assisted works and dare the original creator to defend their rights. They’ll argue that if a machine helped you write it, you don’t own it.

Without a clear standard, every registration becomes a gamble.

⸻

WHAT THE COPYRIGHT OFFICE SHOULD HAVE SAID

Here’s the fix:

If a human controls what gets said and how it gets said—even using AI to get there—they are the author.

It’s that simple. And it tracks with existing law on human–machine collaboration, computer-assisted design, and derivative works. It doesn’t give AI rights. It gives humans credit for using tools well.

The Copyright Office didn’t say that.

Trump didn’t let them say anything else.

So here we are: AI is reshaping creativity. The law is silent. And the people who were supposed to lead just got replaced.

⸻

THE BOTTOM LINE

The Copyright Office tried to map the future of authorship. It failed.

Trump didn’t like the map. So he lit it on fire.

Now we’re all flying blind.

And the fact that no one seems to be talking about it?

That’s not just a media failure. It’s a systems warning.

This won’t stay quiet.

The lawsuits are coming.

The courts will be asked to solve what the Copyright Office couldn’t—and what Congress never tried.

Hope your drafts are saved.

⸻

🤖 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