If you've felt confused about what AI companies can and can't do with your writing, you're not alone, and the confusion isn't your fault. This is one of the fastest-moving legal questions in publishing right now, and even courts are still working it out. Here's the good news up front: the ground is shifting in authors' favor, and understanding why the disagreement exists in the first place makes the whole picture much less overwhelming.

The Actual Disagreement

At the center of this is one legal question: is training an AI model on copyrighted books "fair use," or does it require permission and payment?

AI companies have generally argued yes, it's fair use. Their reasoning goes something like this: when an AI model trains on a book, it isn't storing or reproducing that book, it's learning patterns of language the same way a human writer absorbs style and structure from everything they've ever read. Some judges have been sympathetic to that comparison. One federal judge, ruling in a case involving Meta, compared AI training to the process of teaching schoolchildren to write well, and dismissed concerns that this created unfair competition with the authors those books came from.

Authors and publishers see it very differently, and for good reason. A person learning to write and a company running millions of books through a commercial model to build a product it sells for profit are not the same kind of "reading." The concern isn't philosophical, it's practical: books were taken and used to build billion-dollar products without the authors' knowledge, consent, or compensation. Adding to that, several major cases have revealed that some of these books weren't obtained through legitimate channels at all; they were pulled from pirated "shadow libraries," which shifts the legal conversation from "is learning fair use" to "was this simply theft."

That's the disconnect in a sentence: AI companies are largely arguing about what training does technically. Authors are arguing about what happened to their work practically. Those are two different conversations happening at the same time, which is part of why this has been so hard to resolve.

Why This Is Actually Good News for Authors

Here's where things get encouraging. The legal ground has moved substantially over the past year, and it's moved toward authors.

In 2025, Anthropic agreed to pay authors $1.5 billion to settle claims that it trained its AI on pirated books, the largest copyright settlement in U.S. history. Roughly 500,000 works were covered, and the vast majority of eligible authors filed a claim. That number alone signals something important: when the facts show a company used pirated material, courts and companies are taking authors' claims seriously, and settling for real money rather than fighting to the last dollar.

That settlement has also changed the conversation industry-wide. Legal analysts now expect it to push other AI companies toward settling similar claims rather than risking a trial, because a settlement of that size resets what everyone in the industry expects a losing case to cost. Multiple publishers, including major houses and The New York Times, have ongoing suits against AI companies, and more rulings and potential settlements are expected through the rest of 2026.

At the same time, a real alternative path is emerging: licensing. Rather than fighting in court indefinitely, a growing number of AI companies are striking direct licensing deals with publishers and platforms, paying to use content legitimately instead of defending fair-use claims. It's not a perfect system yet, questions remain about whether independent and smaller authors get a fair seat at that table, but it represents a fundamental shift from "we don't need your permission" to "we'll pay for the right to use this."

What This Means for You Right Now

You don't need a law degree to protect yourself, and you don't need to panic either. A few grounded takeaways:

Your copyright exists the moment you write something down, regardless of what any AI company argues about fair use. That hasn't changed and isn't in dispute in any of these cases.

Courts have also been clear about something authors should find reassuring: AI-generated output itself generally cannot be copyrighted unless a human's creative contribution is present and documented. That protects the value of genuinely human-authored work in a market that's about to be flooded with AI content.

If you're ever unsure whether your specific book was involved in a training dataset connected to a settlement, organizations like the Authors Guild have been tracking these cases closely and are a reliable place to check claim eligibility.

This is a legal and business fight that authors are increasingly winning, one settlement, one lawsuit, one licensing deal at a time. You don't have to fight it alone, and you don't have to fully understand every court filing to know that the industry is moving toward respecting what you make.

Sources:
AI Business, "AI Lawsuits in 2026: Settlements, Licensing Deals, Litigation"
PYMNTS, "US Courts Poised to Shape the Future of AI Copyright Battles in 2026"
Norton Rose Fulbright, "AI in litigation series: An update on AI copyright cases in 2026"
AI Copyright Legal, "AI Copyright Licensing 2026: Publisher Deals Reshaping the Industry"

KL Adams is a literary blogger and fiction writer specializing in dark fantasy, vampire fiction, and paranormal romance. Follow on WordPress, Inkitt (https://www.inkitt.com/KLAdams), Wattpad (https://www.wattpad.com/KLAdams53), and Ream (https://reamstories.com/kladams) for reviews, reading lists, and stories that haunt you long after the last page.

Check out my Gumroad course: https://kladams53.gumroad.com/l/hisznv