Hank's IP Brew — Issue #33

Howdy, y'all.

This week: college jerseys are becoming billboards — and the numbers are bigger than you'd expect. And a federal appeals court just drew a line that every creator who licenses their work needs to understand.

Let's get into it.

In January 2026, the NCAA quietly approved a proposal allowing Division I programs to sell ad space on uniforms, equipment, and apparel for non-championship competitions. Nine months later, the market has arrived — and it's moving faster than anyone predicted.

Here's what happened in a single week in late July, just as college football was getting underway:

Illinois signed a five-year, $30 million deal with Busey Bank covering nine sports — with the bank's logo appearing not just on player jerseys but on coaches' sideline gear as well. Hours later, Ohio State announced a deal with JPMorganChase, brokered by Learfield, worth north of $15 million annually — described at the time as one of the most significant marketing deals in college sports history. That record lasted about six hours. Notre Dame and SoFi then announced a partnership that sources told Sports Business Journal is worth well into the nine figures on a total-value basis, with annual value at or above what Ohio State is getting. The Athletic and Yahoo Sports reported the deal at $18 to $20 million per year over six years, covering all Irish sports.

Notre Dame's deal is currently the largest college jersey patch arrangement on the books. But the market is still early.

What's driving the deals

Learfield, which brokered the Ohio State deal, tracks the college sponsorship market closely. Its 2025-26 NIL Impact report, released this summer, put some numbers on how fast things are moving:

That last number matters for understanding how jersey patches fit into the broader picture. The deals aren't just about the logo on the jersey — the most sophisticated arrangements layer athlete content creation, NIL activations, and school brand-building on top of the patch placement itself. Wisconsin's deal with Culver's, for example, explicitly includes "unique content creation opportunities" and additional NIL components alongside the jersey logo. The patch is the anchor; athlete-facing commercial opportunities are built around it.

The structure of the market

A few patterns are emerging across the deals that have been announced so far.

Most patch partners have regional ties or pre-existing relationships with the school. Busey Bank is an Illinois-area institution. L&N Credit Union already held the naming rights to Louisville's football stadium before adding jersey patches. UPMC was already the presenting sponsor of Pitt women's sports. The patch deals tend to deepen existing relationships rather than introduce entirely new ones — which makes sense from a brand perspective, since jersey placement is most valuable when the audience already associates the brand with the team.

Financial services and healthcare are the two most-represented industries so far. Banks, credit unions, and health systems have regional footprints that map naturally onto fan bases. DoorDash's Stanford deal is notable precisely because it breaks that pattern — the tech company's roots as a Stanford class project give it the institutional connection, but it's the first major tech brand to lead with a patch deal rather than a more traditional sponsorship.

Deal structures almost universally encompass all available sports rather than individual teams, and most include some form of athlete-facing component. Schools are also adding a wrinkle that doesn't show up in the headline numbers: several deals include funds specifically designated for student-athlete benefit, separate from the patch revenue itself, which allows schools to demonstrate athlete investment without those funds counting against the revenue-sharing cap.

What it means for brands and athletes

Jersey patches represent a fundamentally different type of exposure than traditional college sports sponsorship. A logo on a jersey appears in broadcast, streaming, and social media content generated by the school, the athletes, and fans — across every game and every highlight clip, for the full length of the deal. For brands with regional footprints trying to reach the 18-to-34 demographic, the value proposition is significant.

For athletes, the downstream opportunity is in the content layer. Brands that sign patch deals are also structuring NIL activations alongside them — which means athletes at patch-sponsored schools now have a natural pipeline to commercial relationships with brands already invested in their program. The Wisconsin/Culver's content creation component is the template: the brand gets the jersey placement and the athletes get the NIL opportunities, and both sides benefit from the association.

Learfield CEO Cole Gahagan framed the Notre Dame and Ohio State deals this way: "Our biggest brands in college athletics can rival the biggest brands in professional U.S. sports, period." Given that professional leagues have been running jersey patches for years at significant valuations, that comparison is now being tested in real time.

The full list of deals is still growing. Stanford and DoorDash announced their arrangement on September 25 — one of the most recent entries in a market that shows no signs of slowing down before the end of the football season.

On September 16, the Ninth Circuit Court of Appeals issued a ruling in Doe v. GitHub that will matter to every creator who licenses their work and expects attribution in return.

The question before the court: when GitHub Copilot produces code that closely resembles a programmer's copyrighted, openly licensed work — but strips out the attribution notice and license terms that accompanied the original — does that violate the Digital Millennium Copyright Act?

The Ninth Circuit said no. And the reasoning tells you something important about where your legal protections actually end.

What the case was about

The plaintiffs are anonymous programmers who published their code on GitHub under open-source licenses — the kind that permit others to use and build on the code, but require attribution and preservation of the license terms. GitHub Copilot, an AI coding assistant built by GitHub and Microsoft and trained on millions of public GitHub repositories, sometimes produces output that substantially reproduces those programmers' code. But it doesn't include the attribution or license notices that the originals carried.

The programmers sued under Section 1202(b) of the DMCA, which prohibits the intentional removal or alteration of copyright management information — the notices, attribution statements, and license terms attached to a copyrighted work. Their theory: Copilot took their work, stripped the attribution, and reproduced the code without it. That's removal of copyright management information, which the DMCA expressly prohibits.

The district court dismissed the claim. The Ninth Circuit affirmed.

Why the court said no

The panel's reasoning turns on what "removal" means in the statute. Section 1202(b) bars intentionally removing or altering copyright management information — and both of those verbs, the court held, presuppose that the information was attached to an existing work in the first place.

A generative AI system doesn't retrieve copies of your work and strip the attribution off them. It infers statistical patterns from training data and generates new output. That output is a new work — one that never contained your attribution notice to begin with. There's nothing to remove because the information was never there.

The court put it plainly: the provision "reaches affirmative acts against copyright management information attached to a work that already exists." A tool that generates new works that never contained that information in the first place isn't covered.

The panel did clarify one important point: this doesn't mean attribution removal can never be proven. Where an AI's output matches an original so closely — in every respect except the missing notice — a factfinder could infer that removal did occur. The degree of similarity is evidence. But where the output is clearly a new work rather than a copy, the DMCA theory fails.

What this means for creators

If you license your work — code, writing, images, music — under terms that require attribution, this ruling tells you something uncomfortable: the DMCA's attribution-protection provision may not cover what AI does to your work.

The DMCA was written for a world where infringement meant copying an existing work and stripping its metadata. A search engine that copies your page and removes your byline is doing something Section 1202(b) was designed to address. A generative model that learns from your work and produces something new — without ever carrying your byline — is doing something the statute didn't anticipate.

That leaves creators with two alternative paths. The first is straight copyright infringement — arguing that the AI's output is substantially similar enough to your original to constitute copying. That's a harder case to make when the output isn't a literal reproduction, and it's the theory the labels are litigating against Suno and Udio in separate proceedings. The second is contract: if you licensed your work under terms that require attribution, you may have a breach-of-contract claim against whoever used your work without complying. That theory wasn't before the Ninth Circuit in this case, and the contract claims in Doe v. GitHub are still pending in the district court.

The practical takeaway isn't that attribution is unenforceable — it's that you need to understand which legal theory you're relying on. The DMCA's Section 1202(b) is a narrower tool than it looks. Copyright infringement and contract are the stronger hooks for most creators dealing with AI outputs that borrow from their work without credit.

One more note worth flagging: the Ninth Circuit's reasoning applies specifically to generative AI output. It doesn't disturb the broader principle that removing attribution from an actual copy of your work is a DMCA violation. If someone takes your article, strips your byline, and republishes it — that's still covered. The ruling carves out generative output; it doesn't carve out copying.

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.