Authors who sued Anthropic over AI training data are now fighting a second war. This one is against their own publishers and agents. At stake is who actually gets paid when the checks come. According to reports from multiple publishing industry sources, authors are pushing back hard on contractual claims that could redirect settlement money away from the writers who brought the suits in the first place.
What Is Actually Happening
Anthropic, the company behind Claude, faced copyright infringement claims from authors who argued their books were used to train AI models without consent or compensation. A settlement is now in negotiation. But before authors see a dollar, publishers and literary agents are asserting that standard contract language entitles them to a cut.
The Authors Guild, which represents more than 13,000 writers according to its own membership figures, has come out strongly against this interpretation. Authors argue that AI training rights were never contemplated in their original contracts, which means publishers and agents have no legal basis to claim a share.
According to the National Writers Union, many book contracts signed before 2020 contain no language addressing AI training rights at all. The argument from authors is simple: you cannot claim a percentage of rights you were never granted.
This is not a small dispute. Copyright lawsuits against AI companies, including cases against OpenAI, Meta, and Google, have included demands totaling billions of dollars across multiple filings, according to court records reviewed by The New York Times.
The Contrarian Take Nobody Wants to Hear
Here is my read on this situation. Publishers and agents are doing what every middleman does when new money appears on the table. They reach for the closest contract and start reading.
I do not blame them for trying. But I think authors are right to fight it.
The core issue is that this settlement money, if it materializes, exists because authors were harmed specifically as creators. Not as clients. Not as content providers to a distributor. As people whose creative work was scraped and used without permission. The publishers did not write those books. The agents did not write those books. The harm was to the author.
According to a 2024 report from PEN America, more than 90 percent of authors surveyed said they had received no compensation, no notification, and no opt-in opportunity before their works were used in AI training datasets. That is not a publisher problem. That is an author problem. The remedy should flow to the same place.
Compare this to music licensing. When a record label sues over a sample, the label often keeps the bulk of any settlement because the label holds the master rights. Authors who signed away broad rights in their book contracts may find themselves in a similar trap. This is exactly why contract language matters and why authors have historically been in a weaker negotiating position than musicians with strong union representation.
The rich play by different rules here. A debut novelist signed a standard trade deal in 2018 with no AI carve-out has almost no. A bestselling author with a good IP attorney and an updated contract may be in a position to keep every dollar. Same harm, wildly different outcomes based on paperwork signed years before anyone was thinking about large language models.
If you are a creator of any kind, this is your wake-up call. Contracts written before 2023 almost certainly do not protect your AI rights the way you think. That gap is costing people money right now.
For creators building a content business today, tools like InVideo AI let you turn written content into video without handing ownership of your creative output to a platform. Owning your production pipeline matters more in this environment, not less.
What This Means for You
If you are a writer, a creator, or anyone who produces original content, here is what I would do today.
First, pull out every content contract you have signed in the last five years. Look for AI language. If you do not see explicit carve-outs for AI training rights, assume you may have already waived them depending on how broadly the contract defines licensing.
Second, do not assume your agent is protecting these rights automatically. This dispute shows that agents may have competing financial interests when settlement money hits the table. Get clarity in writing before any settlement closes.
Third, if you are creating new content, demand AI training opt-out language in every deal going forward. This is now standard practice for sophisticated creators. If an editor or publisher resists, that tells you something important about how they view your work.
Fourth, diversify your revenue so that no single publisher or platform controls your income. Creators who depend entirely on traditional publishing deals are the most exposed. Creators who have built direct audiences, newsletters, courses, or licensing revenue outside the traditional gatekeepers have options those authors do not.
For indie creators who want to protect their IP while still building a content business, it pays to own your tools outright. Platforms like AppSumo offer lifetime deals on creative software that let you build your production stack without ongoing subscription dependency.
The authors fighting this settlement battle are fighting for everyone who creates. But waiting for courts to fix the problem is a slow strategy. Build your own protections now.
The Bottom Line
Publishers and agents want a cut of Anthropic settlement money. Authors say the contract language does not support that claim. Both sides are right that this fight matters, but only one side wrote the books. Courts will decide who gets paid. In the meantime, every creator operating under a pre-2023 contract is sitting on unexamined legal exposure. The AI companies were not the only ones reading your fine print.
Frequently Asked Questions
What is the Anthropic authors settlement about?
Authors sued Anthropic claiming their copyrighted books were used to train the Claude AI model without permission or payment. A settlement is under negotiation. The dispute is now also about whether publishers and literary agents are entitled to a share of any settlement proceeds under existing contract terms.
Why are authors pushing back on publishers and agents in this settlement?
Authors argue that AI training rights were never included in their original book contracts, which means publishers and agents have no legal basis to claim a percentage. Standard contract clauses were written before AI training was a commercial issue, and authors say those clauses should not be read to cover this new category of rights.
How much money is at stake in AI copyright cases?
Across all pending AI copyright cases, demands from authors, publishers, and news organizations run into billions of dollars collectively, according to court records reviewed by The New York Times. The specific terms of the Anthropic settlement have not been publicly disclosed.
What can creators do to protect their AI rights today?
Review all existing contracts for AI training language and demand explicit opt-out clauses in any new agreements. Work with an IP attorney familiar with AI copyright issues to assess your current exposure. Building direct revenue streams outside traditional publishing reduces your dependence on contracts that may not protect your interests.
Does this authors vs. Anthropic dispute affect other creators beyond book authors?
Yes. Similar legal questions apply to musicians, visual artists, journalists, and any creator whose work was publicly available online before 2023. The contract and rights questions being tested in the authors settlement will set precedents that affect how all creative IP is handled in AI training disputes going forward.
