Anthropic just wrote a check for $1.5 billion to settle a class-action lawsuit with book authors, and a federal judge in San Francisco nodded it through. The number is staggering—the largest copyright settlement in class action history. But before anyone declares a victory for creators everywhere, look at what the money actually bought. The payout covers Anthropic’s use of pirated books from shadowy databases like LibGen and PiLiMi, scooped up in 2021 and 2022. Of the nearly half a million works listed in the suit, over 91 percent were claimed, netting roughly $3,000 per title. That’s four times the statutory minimum. Anthropic also has to nuke the stolen files.
So, pirates lose, authors get paid. Case closed? Hardly. The real fight—whether training AI models on copyrighted work is legal in the first place—remained completely untouched by this settlement. Judge Alsup has already tipped his hand on that question, calling AI training on legally obtained books “transformative—spectacularly so” and a clear case of fair use. What this settlement does is surgically separate the crime of downloading from shady sites from the act of feeding data into a neural network.
For AI labs, this is the best billion-and-a-half dollars they could possibly spend. It buys legal clarity without setting a precedent that guts their entire training pipeline. The murky part—whether scraping the open web without explicit consent counts as “legally obtained”—is still floating out there, unresolved. But for now, the core practice of hoovering up public data to build models stands. Anthropic pays for its shoplifting. The AI industry keeps its factory.
I’d call it a pyrrhic victory for authors if the check weren’t so large. But make no mistake: the settlement carves out AI outputs that reproduce original works and future conduct, leaving room for more litigation. This isn’t the end. It’s a very expensive comma.
