Why AI Companies Choose Lawsuits Over Licensing

When you strip away the lawsuits, the cease-and-desists, and the billion-dollar settlements, the book industry and AI companies are locked in a problem of psychology, not economics. And Trip Adler knows this because he’s tried to sell them a cheaper way out.

Trip Adler is building Created by Humans — a licensing marketplace for AI rights. He previously built Scribd over 16 years, the subscription platform that finally convinced publishers that digital books were worth protecting. Now he’s attempting something harder: convincing AI companies that paying authors and publishers is smarter than fighting them in court.

The surprising part? The market logic doesn’t work. Companies could license books for less than they’ll spend on litigation. But they don’t. Understanding why reveals something fundamental about how companies make decisions under uncertainty.

The Economics Don’t Add Up

Here’s what you’d think would happen: AI companies see the legal risk, see the licensing option, calculate the cost of litigation, and choose the cheaper path. Instead, they’re choosing expensive lawsuits.

“There’s been some AI companies that really believe in this model,” Trip explains. “Not everybody just wants to get data without permission.” But here’s the gap: the companies that say they believe in licensing aren’t the ones licensing at scale. The ones licensing are smaller startups and specialized teams. The big players? They’ve already trained on pirated data. Now they’re in a different calculation entirely.

Trip’s observation cuts to something most founders miss: once you’ve already used the data, licensing becomes retroactive liability management, not forward-looking cost avoidance. “A lot of these companies have already trained on the books using pirated versions,” he notes. “We have this challenge of like, well, you already have the books, but let’s try to do this in a more legitimate way.”

That’s the real battle. It’s not about the business logic of going forward. It’s about admitting past mistakes and committing to doing it differently. Lawsuits, by comparison, let you stay in argument mode indefinitely.

The Multi-Rights Complexity Nobody Wants to Solve

Even when Trip finds AI companies interested in licensing, there’s another wall: the rights problem.

A single bestselling book isn’t owned by one person. There’s the author, the co-authors, the publisher, the literary agent, sometimes multiple publishers across territories. When an AI company wants to license a million books, they’re not negotiating one deal — they’re negotiating millions.

“When it comes to trade books, the rights get pretty complicated and there’s multiple rights holders,” Trip explains. “For a particular book, you typically need to get sign-off from the author, maybe the co-author, the publisher, sometimes there’s multiple publishers involved, there’s often a literary agent involved. So there’s a lot of rights holders that need to be cleared for a particular book.”

This complexity is why most AI companies didn’t even try. “They looked at how complicated the rights were and they’re like, this would be impossible to clear the rights for, because they all needed a lot of books, like a million books, it was just impossible to clear the rights for the books.”

So Created by Humans had to solve what the entire industry thought was unsolvable: a platform that let authors sign up individually, select which rights they wanted to license, and automatically split revenue across all stakeholders. “We’ve built a way to kind of scalably get these rights,” Trip says. But scaling toward zero traction is still scaling toward zero.

The Market Says No — Now What?

The deeper problem emerges when you talk to AI companies about what would actually move them. It’s not about price. It’s about what you’re admitting.

Trip has done deals. Some training deals, some reference deals, some transformative. But the volume is nowhere near what it needs to be to move the market. The lawsuits, by contrast, are happening at scale. Britannica suing OpenAI. Authors suing Microsoft. The cost of litigation is real, but it’s also diffused — it lives in legal departments, not product budgets. Licensing costs are immediate and visible.

There’s also something else: lawsuits maintain the fiction that you’re fighting about principles, not money. Licensing admits you’re negotiating a price for something you’re admitting you wanted to take. For a CEO trying to manage investor relations and employee morale, that’s harder positioning than “we’re confident our use is fair.”

Trip’s candor about this gap is what makes him credible. He’s not pretending the market has changed. “Most AI companies are choosing litigation over licensing even when licensing is cheaper,” he acknowledges. But he’s also betting that the courts will eventually force the conversation.

“I think once these lawsuits mature, that’ll also be another catalyst for this to pick up even more,” Trip says. The moment a court rules that a company can’t use unpaid data without permission, licensing stops being optional and becomes table stakes.

What Actually Moves the Needle

Trip’s success has come not from convincing big labs, but from finding the edge cases: startups that want to do it right from the beginning, specialized companies building AI tools for verticals where licensing is already expected, even a government agency.

“There’s been a mix,” he explains. “Some of the big guys, we’ve done deals with the big AI companies, the names you would know. And there’s also a lot of startups. There’s a lot of startups that are interested in this, which are just some earlier stage version of bigger AI companies.”

The pattern is clear: startups and scrappy teams will license. Well-funded labs with legal teams will litigate. The question for founders building AI is which path to choose now, before you’re forced to retroactively negotiate.

“If you’re a startup founder listening to this and you’re looking for licensed data, I mean, we’d love to talk,” Trip says. “Because I think it’s the right way to do things. And it’s a great way to just get visibility and do things the right way.”

The real cost of litigation isn’t the settlement you pay. It’s the years spent in legal limbo while your product roadmap waits for a court decision.

FAQ

Why would a company choose an expensive lawsuit over cheaper licensing?

Once a company has already trained on unlicensed data, litigation becomes about managing past decisions, not making rational forward choices. Licensing becomes an admission of wrongdoing. Litigation lets you claim fair use is still an open question.

What makes book licensing so hard compared to music or other media?

Books have fragmented rights ownership. A single book might involve authors, co-authors, publishers in different territories, literary agents, and co-authors. An AI company wanting to license a million books faces millions of separate negotiation points, which publishers saw as impossible until Created by Humans built tooling to automate the process.

Are there AI companies actually licensing books today?

Yes, but mostly smaller companies and specialized players. Some major labs have done deals, but the volume is far below litigation costs. Most licensing deals today are reference rights (RAG retrieval) rather than training rights, which have proven more acceptable to authors.

What would force AI companies to license instead of litigate?

Court rulings that establish unpaid data use isn’t fair use. Licensing is expected to accelerate significantly once lawsuits mature and create legal precedent that licensing is required, not optional.

How much does licensing cost for AI training?

Every deal is different, and the market is still discovering pricing. Trip structures deals as usage-based fees, upfront payments, or revenue share agreements depending on what works for both sides. There’s no standard yet — the pricing is being discovered deal by deal.

Can startup founders afford to license data for AI?

Yes. Trip structures deals to work with companies of all sizes. Smaller teams might use entirely usage-based pricing that scales with their actual AI usage, while larger companies might pay upfront fees or revenue splits.

Is litigation uncertainty actually expensive compared to licensing?

Yes. A single lawsuit with major publishers can cost millions in legal fees, plus settlement amounts in the hundreds of millions (like Anthropic’s $1.5B author settlement). Licensing ahead of time would be significantly cheaper in most scenarios.

Who decides which rights to license — authors, publishers, or both?

Created by Humans’ model lets authors and publishers claim their works and decide which rights they’ll license. For books with multiple rights holders, the platform handles revenue splitting. This is different from most AI marketplaces, which negotiate directly with publishers only.

What’s the difference between training, reference, and transformative rights in AI licensing?

Training rights let AI companies use books to train models. Reference rights let AI systems cite and link back to books (like RAG). Transformative rights let AI create derivatives or modifications based on the source material. Each has different legal and business implications.

Will this ever be as simple as music licensing?

No — books are more complex than music. A song has one copyright holder (usually). A book might have five or more. But Trip is building infrastructure to make complex multi-party deals work at scale, similar to how YouTube handles copyright for millions of videos with split ownership.