📚💰 Anthropic's $1.5B "Fine" Was An Amazing Outcome for Them

In partnership with

Anthropic agreed to pay $1.5 billion to settle a class-action lawsuit from book authors who claimed the company pirated their works to train Claude. The headlines write themselves: "AI company pays massive fine for copyright infringement." But if you squint at this story, something more interesting emerges.

The federal judge's ruling established something that will matter far more than the settlement amount: training AI chatbots on copyrighted books is perfectly legal as long as you acquire the books legally. What Anthropic got dinged for was downloading 7 million books from piracy websites like Library Genesis instead of buying them like a normal person.

The math here is... telling. Anthropic will pay authors roughly $3,000 per pirated book. They could have bought each book legally for about $15. So we're talking about a 200x price difference, which creates some fairly obvious incentives going forward.

For AI companies, this represents a clear win. They've converted what could have been expensive ongoing costs (licensing fees) into manageable one-time expenses (legal acquisition).

For $1.5 billion, Anthropic just established legal precedent that makes all future content acquisition dramatically cheaper. Whether they meant to or not is another question, but that doesn’t really matter.

Paying For Content Licensing Dream Dies Quietly

Before this ruling, plenty of well-intention people argued that AI companies should pay creators ongoing royalties, much like how YouTube shares revenue with rights holders when copyrighted music appears in videos. The economics seemed to work: YouTube prints money, so why not AI?

Well, the court just answered that question. The judge explicitly ruled that training on copyrighted material is legal with proper acquisition. No ongoing compensation required, thank you very much. There's no "YouTube for AI" coming because the legal system just said AI companies don't need one.

This extends beyond books, too. AI companies might still cut deals with major news publishers for current information (breaking news doesn't train itself), but they won't need licensing agreements with every content creator. There are diminishing returns here, after deals with the top 10 news sources, what exactly does the 11th publisher add?

The Boom in Expert Labelled Data

While everyone argues about copyright law, something else is happening. Companies like Scale AI, Handshake, Surge, and Mercor have built businesses around providing expert human feedback to train and fine-tune AI models. These aren't your typical data labeling operations, they recruit domain specialists in coding, law, finance, medicine, and consulting to teach AI systems nuanced reasoning in their fields.

The numbers are pretty wild. These expert feedback companies have combined valuations north of $70 billion. Google Cloud, Microsoft Azure and Amazon AWS in 2018 (12 years after the start of AWS) had a combined market cap of $444 billion. That means in just 3-5 years, the expert data labeling market has reached almost 20% the size of the cloud computing market after 12 years.

AI companies are moving from "scrape everything" to "curate with specialists." They're still hoovering up massive datasets for pre-training (this hasn't changed), but the marginal value increasingly comes from expert human feedback during fine-tuning. Domain experts are pulling $150-200 per hour to train and evaluate AI models.

The shift toward specialized expertise is accelerating. Elon Musk's xAI just laid off 500 generalist data labelers while announcing plans to 10x their specialist tutors & AI researchers at Meta are now asking to use competing services like Surge and Mercor because Scale's quality isn’t good enough.

Generic human feedback is getting commoditized, but specialized expertise commands serious premiums.

Arbitrage Opportunities Abound

That 200x cost difference between legal acquisition and piracy penalties creates some obvious business opportunities. If there's such a ripe and growing market for expert data labelers, there could also be a business that helps labs get access to data in books legally now that doing so is 200x cheaper. Expect new companies to emerge that legally buy books, scan them, and license clean datasets to AI labs. Think Scale AI, but for content acquisition instead of data labeling. The business model is simple: buy a book, rip out the pages, scan them, and sell clean datasets to AI companies.

If AI labs pivot to ebooks instead of physical scanning, Amazon gets a small windfall. Even if all major labs buy 7 million additional books each, that's about $420 million in extra revenue against Amazon's $17 billion annual ebook business. Nice, but not exactly transformational.

Historical Echoes

This playbook has been run before. Google faced similar lawsuits over Google Books from 2004-2013, eventually winning on fair use grounds.

Anthropic's $1.5 billion serves a similar function. They've established that written knowledge can be legally acquired and processed for AI training. And now every other AI company benefits from this precedent, Anthropic just had to pay some legal fees.

How 433 Investors Unlocked 400X Return Potential

Institutional investors back startups to unlock outsized returns. Regular investors have to wait. But not anymore. Thanks to regulatory updates, some companies are doing things differently.

Take Revolut. In 2016, 433 regular people invested an average of $2,730. Today? They got a 400X buyout offer from the company, as Revolut’s valuation increased 89,900% in the same timeframe.

No wonder 10K+ everyday people are taking the chance on Pacaso.

Founded by a former Zillow exec, Pacaso’s co-ownership tech reshapes the $1.3T vacation home market. They’ve earned $110M+ in gross profit to date, including 41% YoY growth in 2024 alone. They even reserved the Nasdaq ticker PCSO.

The same institutional investors behind Uber, Venmo, and eBay backed Pacaso. And you can join them. But not for long. Pacaso’s investment opportunity ends September 18.

Invest While You Still Can

_{Paid advertisement for Pacaso’s Regulation A offering. Read the offering circular at invest.pacaso.com. Reserving a ticker symbol is not a guarantee that the company will go public. Listing on the NASDAQ is subject to approvals.}