Authors who sued Anthropic over AI training data are now fighting a second battle. Publishers and literary agents want a share of any settlement payout. I’ve watched a lot of money disputes play out in tech and finance. This one sets the rules for who actually owns creative work in the AI era, and most creators are completely unprepared for what comes next.
What Is Actually Happening
In 2023, authors filed suit against Anthropic, claiming the company used their copyrighted books to train Claude without permission or payment. The lawsuit moved slowly, as copyright cases do. By 2026, settlement talks are real and the numbers are significant enough that everyone wants a piece.
The problem is that publishers and literary agents are now arguing they deserve a portion of any payout. Their argument rests on old contract language. Standard literary agent agreements entitle agents to 15% of income derived from an author’s work, according to the Association of Authors’ Representatives. Publishers are making similar arguments, pointing to broad “derivative works” and “subsidiary rights” clauses buried in contracts signed years or even decades ago.
According to the Authors Guild, AI companies ingested millions of copyrighted books without author consent during the development of their large language models. The Authors Guild’s position is firm: settlement compensation belongs to the creators whose work was used, not to the businesses that distribute that work.
Authors are pushing back just as firmly. Their argument is simple. AI training rights were never part of what they negotiated or signed away. Publishers and agents are trying to collect on rights they never represented and never paid for.
The Money Grab Nobody Is Talking About
Here is what most coverage misses. This fight is not really about Anthropic. It is about contract language written long before anyone imagined that feeding books into a neural network could become a revenue stream.
Publishers and agents are using vague old contract language to claim rights they never negotiated for. That is the same playbook major record labels ran when streaming royalties started flowing. Find language broad enough to cover new money. Move fast. Let the creator fight you in court if they want their share back.
According to a 2024 Authors Guild survey, the vast majority of traditional publishing contracts contain no specific language addressing AI training rights. That gap did not appear by accident. Vague contracts benefit the party with more lawyers and more money to litigate. That party is almost never the author.
I think about this the same way I think about any wealth transfer. The people who own the asset cleanly, with no ambiguous middlemen attached, are the ones who collect when the asset generates new value. Authors who retained their IP inside a clean legal structure will have a stronger claim than authors who signed broad rights away to a publisher in 2005 and never gave it another thought.
When new deals come your way now, don’t sign anything without explicitly carving out AI training rights. A platform like signNow lets you review, annotate, and digitally sign contracts before you commit, so you catch the missing language before it costs you a settlement check years down the road.
What This Means for Creators
This dispute is a warning shot for every person who produces original work: writers, developers, designers, educators, consultants. Here is what I would do right now.
Pull every contract you’ve signed in the last ten years. Look for AI training rights language. Most contracts signed before 2023 have none. That ambiguity will play out differently in different courts. Some judges will read old “subsidiary rights” language broadly enough to cover AI uses. Others won’t. You don’t want to find out which way your contract reads after a settlement offer lands on the table.
Stop signing contracts that leave AI rights unaddressed. Any agent or publisher who resists adding explicit AI training language to a new contract is telling you something important about how they plan to handle the next dispute.
Think about your ownership structure. IP attorneys consistently advise that creators who hold their intellectual property inside a separate business entity have cleaner standing in licensing disputes than sole proprietors operating under their personal name. Forming an LLC creates a legal wall between your creative output and the claims of distributors who follow the money after the fact. Inc Authority offers free LLC formation, and putting your IP inside that entity before you sign your next deal is one of the most straightforward moves a creator can make right now.
The authors fighting this battle with Anthropic did everything right by filing the original suit. But the lesson from their second fight is that winning in court is only half the work. Owning the outcome requires owning the structure underneath it before the money appears.
The Bottom Line
Authors went to court to protect their work. Now the people who were supposed to represent them are lining up to take a percentage of the win. This will not be the last time this happens. Every creative industry is heading toward the same argument. The creators who come out ahead will be the ones who owned their IP cleanly before the money showed up. Everyone else will be negotiating from the back of the line.
Frequently Asked Questions
What is the Anthropic author settlement dispute about?
Authors sued Anthropic for using their copyrighted books to train Claude without permission or compensation. Now that settlement talks are underway, publishers and literary agents are claiming they deserve a share of any payout. Authors are pushing back, arguing that AI training rights were never part of what they originally signed away.
Can publishers legally claim part of an AI copyright settlement?
It depends entirely on the contract language. Standard publishing agreements often include broad “subsidiary rights” and “derivative works” clauses. Publishers argue those clauses cover AI training compensation. Authors argue they do not. Courts will decide this on a deal-by-deal basis, and outcomes will vary.
What should authors do to protect themselves from similar claims?
Review every existing contract for AI rights language. Add explicit AI training exclusions to any new contract before signing. Consider holding your intellectual property inside a business entity, which creates cleaner ownership and stronger standing in future disputes.
How do literary agents fit into AI copyright disputes?
Literary agents typically receive 15% of income from an author’s work, according to standard industry agreements published by the Association of Authors’ Representatives. Agents are arguing that any settlement income from AI copyright claims falls under that commission structure. Authors dispute this, saying agents never negotiated or protected AI rights in the first place.
Will this change how publishing contracts are written going forward?
Yes, and it already is. The Authors Guild and other advocacy groups are pushing for AI training rights to be explicitly listed and retained by authors in all new contracts. Publishers who resist adding this language are setting up the conditions for more disputes exactly like this one.
