The blank book protest is also a commentary on how AI companies value creativity. They need authors' work to train models—the data doesn't appear from nowhere. But they want it for free, arguing that the value is in the AI system they built, not the content they trained it on. It's like arguing that a music streaming service shouldn't pay royalties because the value is in the algorithm, not the songs.

From a technical perspective, large language models absolutely require massive text datasets. You can't train GPT-scale models without billions of words of human-written text. The question isn't whether authors' work has value to AI companies—clearly it does. The question is whether authors should have any say in how that value is extracted.

The copyright law angle is genuinely complicated. Fair use allows certain uses without permission, including transformative uses. AI companies argue that training a model is transformative because the output isn't a copy of the input. Authors argue that the value extracted during training still requires compensation, even if the final output is transformative.

But the legal complexity doesn't justify the approach. Just because something might be legal doesn't mean it's right. AI companies could have built opt-in systems. They could have approached authors early and negotiated reasonable terms. Instead, they took everything first and are arguing about legality later.

What frustrates me most is the asymmetry. Authors spent years writing books. AI companies spent months scraping and training. The authors own the copyright but have little leverage. The AI companies own the models and have billions in funding. The power imbalance means consent is functionally optional.

The empty book won't stop AI training. Companies can still use the millions of other books they've already scraped. But it's a statement: we didn't consent, we don't consent, and we're organized enough to fight back. Whether that matters legally will depend on courts. Whether it matters practically depends on whether AI companies decide consent is worth the trouble.

The technology is impressive—language models trained on books really can write coherently and answer questions. But the question isn't whether the technology works. The question is whether we're okay with an industry built on taking first and asking permission never.

Authors aren't asking for AI to stop existing. They're asking to be part of the conversation about how their work is used. The fact that they had to publish an empty book to make that point tells you how broken the current system is.