Hank's IP Brew — Issue #27

Howdy, y'all.

This week: a federal judge just signed off on a $1.5 billion settlement — the largest copyright recovery in history — against an AI company that trained on pirated books. And Disney just did something it has never done before: handed creators licensed access to Marvel, Star Wars, and Pixar IP to make videos on TikTok. Both stories are about the same underlying shift — the lines around AI and creator IP are being drawn, one case and one deal at a time.

Let's get into it.

On July 20, Judge Araceli Martínez-Olguín of the U.S. District Court for the Northern District of California granted final approval to the largest copyright settlement in recorded history. The case: Bartz v. Anthropic. The amount: $1.5 billion. The underlying conduct: Anthropic trained its Claude AI models on pirated books sourced from Library Genesis and Pirate Library Mirror — shadow libraries containing hundreds of thousands of copyrighted works downloaded without authorization.

Over 91 percent of eligible class members — authors whose works appeared on the list of pirated titles — submitted claims. The per-work payout is approximately $3,000, which the court noted is four times the statutory minimum for ordinary copyright infringement, and vastly more than the maximum proposed in the Google Books settlement that was rejected in 2011.

Beyond the money, Anthropic is required to destroy all original files sourced from those pirate libraries, along with any copies derived from them.

The ruling that made the settlement possible

To understand what this settlement means, you have to understand the ruling that preceded it. In June 2025, Judge William Alsup drew a line that will define AI copyright law for years: AI companies can use legally acquired works for training. They cannot use pirated ones.

Judge Alsup's reasoning was direct: "Anthropic had no entitlement to use pirated copies for a central library. Creating a permanent, general-purpose library was not itself a fair use." He distinguished between a company that purchases physical books and digitizes them for storage — lawful, because the works were purchased — and a company that downloads the same books for free from a pirate site — not lawful, because those works could have been acquired through legitimate channels.

That distinction — lawfully acquired training data versus piracy-sourced training data — is now the foundational rule in U.S. AI copyright law. Fair use may protect AI training on legitimately obtained works. It does not protect training on stolen ones.

What the settlement does and doesn't cover

The release is deliberately narrow, and that narrowness matters. Class members released only claims related to Anthropic's past acquisition and use of their works — the "inputs" — through August 25, 2025. Output claims are not released. Future conduct claims are not released. Authors whose works don't appear on the Works List are unaffected entirely.

The Authors Guild, which helped compile contact information for the class notice campaign, summarized it plainly: authors "retain their rights to sue for future misconduct."

That means the settlement is an endpoint for the piracy chapter of Anthropic's history, not a general license for AI training going forward. Anthropic still faces potential liability for anything it does after August 2025.

What this means for writers and authors

If you are a writer and your books were published before August 2025, there is a real possibility they were used to train AI models — by Anthropic or by others. The claim window for this settlement has closed, but the legal framework it established is now available to every author in future cases.

The practical implication is straightforward: AI companies that trained on piracy-sourced data have significant legal exposure. AI companies that trained on lawfully acquired data have a stronger fair use argument — though that question remains contested and unresolved in other pending cases. The difference between those two categories is not academic; it is, apparently, $1.5 billion.

For creators in other domains watching this case — visual artists, musicians, filmmakers — the Bartz ruling is the clearest signal yet that provenance of training data matters enormously. "We trained on data we found online" is not a safe harbor if that data was stolen.

On August 5, Disney announced a partnership with TikTok that does something the company has almost never done before: grants creators sanctioned, licensed access to IP from its entire portfolio — Marvel, Star Wars, Pixar, FX, and hundreds of other titles — to make short-form videos.

Creators who opt in can use characters, scenes, and assets from Disney's catalog. Their videos will appear on both TikTok and Disney's new vertical video feed on Disney+, called Verts. Participating creators get additional perks: increased visibility on both platforms, exclusive event access, and what Disney describes as "career development pathways."

For context on why this is significant: Disney has historically been among the most aggressive IP enforcers in the entertainment industry. The company's track record on fan content has ranged from tolerant to litigious depending on the property, the platform, and the commercial context. A formal opt-in program granting licensed access to the full portfolio is a meaningful strategic shift.

What the license actually means — and what it doesn't say

Here's where it gets interesting for anyone thinking about participating.

The announcement confirms that creators get licensed access to Disney assets for fan-created short-form videos distributed on TikTok and Disney+. What it does not clearly address — and what Disney has not yet publicly clarified — is who owns the resulting video.

That distinction matters enormously. There are three possible ownership structures for a derivative work created under a license like this:

Until Disney publishes the actual license terms for the program — which it hadn't done as of this writing — creators should not assume they own the videos they make under this partnership.

The AI question nobody has answered

The Verge noted something important in its coverage: Disney's announcement says nothing about excluding AI tools from the creator workflow. Disney had previously been negotiating a separate deal with OpenAI's Sora video generator that would have allowed Sora to train on Disney IP in exchange for hosting Sora-made videos on Disney+. That deal fell apart when Sora was discontinued. But the architecture — Disney IP + AI generation — clearly has Disney's interest.

If you use an AI tool to help create your fan video under this program, are you within the license terms? Is your AI-assisted video eligible for Disney+ distribution? Does the license require human authorship? These questions are unanswered, and they matter before you build a content strategy around this program.

The bigger picture

Disney doing this at all is the real story. Major IP holders have spent years trying to suppress, ignore, or litigate fan creativity rather than channel it. A formal licensing program that puts fan-made content on Disney+ alongside official titles is a structural acknowledgment that fan creativity has commercial value — and that controlled access beats uncontrolled enforcement.

Other IP holders will watch this closely. If Disney's program generates meaningful engagement and creator loyalty without diluting brand value, expect more structured fan-creation licensing from other studios, sports leagues, and major brands. The question of who owns derivative fan content — and what platforms owe creators for hosting it — is going to become a live issue fast.

For now: if you want to participate, wait for the full license terms before building anything on this. Understand exactly what rights you're granting and what you're getting. The opportunity may be real; the fine print will tell you whether it's worth it.

See you next time,

Hank

P.S. If someone forwarded this to you and you find it useful, subscribe at newsletter.creatoripacademy.com. No spam, no filler — just IP news that actually matters.

About Hank's IP Brew

Creator IP Academy helps creators understand and protect their intellectual property. Got a question? Reply to this email.