Author Help (@office.gts.authorhelp.uk.ap.brid.gy)

The Anthropic Class Action Settlement: What You Need to Know Right Now One of the most urgent issues confronting writers and other creators right now is the use of copyrighted material for generative AI training. The large language models that power chatbots like OpenAI’s ChatGPT and Anthropic’s Claude require “training” via the ingestion of vast amounts of text, images, and other materials scraped from the internet or incorporated into databases created by AI companies. Much of this material is protected by copyright. For the most part, AI companies have not sought permission from rights holders for exploiting their work in this way (nor did creators even start discovering the extent of the companies’ use of their content until a few years ago). They claim permission isn’t needed because AI training falls under the definition of fair use—a limited and transformative use of the material that does not require the copyright owner’s agreement (or remuneration). Not surprisingly, most copyright holders disagree. AI training is a new right under copyright law, they argue, and therefore cannot be exploited without explicit authorization from the copyright holder. This fundamental dispute has set the stage for a bonanza of lawsuits. According to this master list of legal actions over AI training, there are currently at least 51 copyright lawsuits against AI companies over AI training, with, no doubt, many more to come. While a handful have been voluntarily dismissed, most are still pending. There is, however one that has been resolved: Bartz v. Anthropic. ### **BARTZ v. ANTHROPIC: THE LAWSUIT** A year ago, a group of authors filed a copyright infringement class action suit against AI company Anthropic over its creation of an enormous library of digitized books to train its Claude chatbot. (You can see the original complaint here.) In addition to buying and scanning physical books for inclusion, Anthropic also downloaded millions of books and other works that had been illegally uploaded to pirate sites LibGen & PiLiMi (LibGen is pirate site Library Genesis; PiLiMi is pirate site Pirate Library Mirror)—all without permission from the works’ creators. (Are your works part of the pirate databases? There’s no way to check PiLiMi, but you can use this resource to search LibGen.) This summer, the judge in the case issued a mixed ruling, determining that while Anthropic’s digitzation of properly acquired physical books, and its employment of that digitized library for AI training, did indeed qualify as fair use, its downloading of pirated works was a massive infringement of copyright that justified a trial. In July, the judge certified a class of authors whose works were included in the two pirate databases downloaded by Anthropic. Here’s the judge’s definition of the class: > _All beneficial or legal copyright owners of the exclusive right to reproduce copies of any book in the versions of LibGen or PiLiMi downloaded by Anthropic. “Book” refers to any work possessing an ISBN or ASIN which was registered with the United States Copyright Office within five years of the work’s publication and which was registered with the United States Copyright Office before being downloaded by Anthropic, or within three months of publication._ In other words, the class is limited to only those works with an ISBN or ASIN whose copyright was properly registered, and includes not just the authors of those works, but their publishers. (As the Authors Guild explains: “under copyright law, where the author grants the publisher an exclusive license to publish the book in exchange for percentage [of] royalties, the publisher is deemed the legal owner of the publication rights granted to it, and the author is deemed the ‘beneficial’ owner of that right”.) Together, the two pirate databases incorporated an estimated _seven million_ works. Under US copyright law, the maximum in statutory damages that can be collected for willful copyright infringement is $150,000 per work. Even if only a fraction of the works in the pirate databases fit the judge’s definition of the class, it’s clear that Anthropic was looking at a truly catastrophic payout if they lost at trial. Unsurprisingly, they chose to settle. And if your work was part of one of the pirate databases, it’s possible you may benefit. ### **THE BIGGEST COPYRIGHT SETTLEMENT IN HISTORY** Under the terms of the settlement (you can see the motion for preliminary approval, which includes the details of the settlement, here), Anthropic will pay a _minimum_ of $1.5 billion (yes, _billion_) into a settlement fund. Once the settlement receives final approval, the money will be distributed among rightsholders on a pro-rata basis. Right now, the payout is estimated at just over $3,000 per work, based on a total of 482,460 eligible works on the master list compiled by plaintiffs’ lawyers. That amount won’t go lower, but it could wind up being higher, depending on how long the settlement process drags on (the settlement escrow account will accumulate interest) and how many eligible claimants come forward (plaintiffs’ lawyers are working to find and notify every eligible claimant, but inevitably some will fall through the cracks). $3,000 per work, while nowhere near the statutory maximum for willful copyright infringement (and I don’t think that anyone could plausibly argue that Anthropic’s infringement wasn’t willful) is a good deal of money, especially for authors with multiple works included in the settlement. You won’t receive that whole amount, though. Attorneys’ fees, estimated at 25%, must be deducted, along with the expense of administering the settlement. If you’re the sole legal owner–i.e., your contract has expired, you’ve reverted rights, or you’ve self-published–you get 100% of what remains. If there’s an exclusive contract currently in force, your publisher is considered an additional legal owner/rightsholder and you must share the payout with them. The default split proposed by plaintiffs is 50/50, based on the division of infringement lawsuit proceeds that’s standard in publishing contracts, but it’s non-mandatory, and you’re free to make a case for a different split if you have documentation to support it. The master list of eligible works is available at the official settlement website. There’s a lookup function that allows you to search on your name, your work’s title, the publisher, or the ISBN/ASIN to find out if any of your writings are included. If your work is eligible, you’ll need to file a claim. Official notices will be going out to claimants in the coming weeks, but if you’ve confirmed that your work is on the master list, you don’t need to wait to receive it. For sole legal owners, there’s a short claim form that can be submitted online. A longer form, which you’ll need to use if there are rightsholders besides yourself or you want to challenge the default income split, can be downloaded and submitted by postal mail. The Authors Guild has an excellent step-by-step guide to filling out claim forms, and Writer Beware’s sponsor, SFWA, will soon post a claim form FAQ to its Anthropic resource. The settlement website’s own FAQ.is also very helpful. Claims aren’t due until March 23, 2026, so you have plenty of time to file. If, on the other hand, you want to opt out of the settlement, and thus preserve your ability to separately sue Anthropic, you must do so by January 7, 2026. ### **MANY AUTHORS WILL BENEFIT, BUT MANY ARE LEFT OUT** You may be wondering why, if the pirate databases included seven million works, fewer than 500,000 are actually part of the settlement. Or why, if you found your work in the LibGen database, it doesn’t appear on the settlement’s master list. The master list includes only works that meet the class definition: works with ISBNs or ASINs whose copyright registrations were made before Anthropic’s downloads and within five years of publication. Per the settlement website FAQ, most works in the pirate databases did not meet these requirements. About 40% were duplicates. Other files were corrupted or empty, or were added after the cutoff date of August 2022 (the date of Anthropic’s final PiLiMi downloads), or had registrations made later than five years post-pub. And millions of works didn’t have a copyright registration at all. Although worldwide copyright law ensures that creators have copyright ownership the instant they fix the work in tangible form, the United States is unique in making the extra step of registration with the US Copyright Office a pre-requisite for defending copyright in court. There’s no such requirement in Europe–so many of the millions of non-English-language works included in the pirate databases weren’t registered. Additionally, the settlement has exposed a problem no one suspected existed: even where publishing contracts required the publisher to register copyright on the author’s behalf, many publishers simply…didn’t. You may, therefore, be locked out of the settlement thanks to the peculiarities of US copyright law or your publisher’s failure to honor its obligations, even though Anthropic violated your copyrights just as thoroughly as those of authors who do have registrations. This gigantic settlement, in other words, has significant unfairness at its heart. Also important: the redress on infringement represented by the settlement is not for AI training, but rather for piracy. The judge in the case penalized Anthropic only for its use of pirated material; for its use of _legally_ acquired material, he ruled in its favor, agreeing with its claim that this qualified as fair use. So althought the settlement possibly sets a precedent on the _source_ of materials used for AI training (there’s already another AI training lawsuit based on the use of pirated works)–on the training itself, Anthropic won its case. It’s early days in the battle over AI training and copyright, but if such precedents keep stacking up, writers will experience yet another narrowing of their rights under copyright, in a world where the pressure of technological development is steadily eroding the protections of copyright law. ### THIS MAY BE A LONG PROCESS Payouts won’t begin until after the settlement is finalized. Currently, the hearing on that is scheduled for April 23, 2026, but that may well change, especially if there are substantial objections to the settlement. There are also many ambiguities that have yet to be addressed. What about contributors to anthologies protected by collective copyrights, for example—how will they (or will they) share in the payout? What if there are errors in the master works list, such as the ISBN of a re-published book not corresponding to the edition that was registered? What if the work is a collaboration with two or more authors and copyright ownership isn’t clear? These and other questions will hopefully be resolved and added to the settlement website FAQ, but that too will take time. Fortunately, with claims not due until March of next year, you don’t have to rush to file. You can also contact the Settlement Administrator with questions and concerns, at 877-206-2314 or [email protected]. All other issues aside, this is a historic copyright recovery settlement—the largest ever—and it is your best, and probably only, opportunity to get at least some redress for the pirating of your work. I urge every author who’s eligible to take part, and spread the word so that as many writers as possible know about this opportunity. Authors’ organizations are following the situation closely, including the Authors Guild and Writer Beware’s sponsor, SFWA. Look for updates on those organizations’ websites, and also for updates on the Writer Beware blog. ### **RESOURCES** The official settlement website includes general information, a works list search function, a detailed and extensive FAQ, a list of key dates, and a claim portal. Contact the Settlement Administrator: 877-206-2314 or [email protected]. Look up your copyright registration in the US Copyright Office’s copyright registration database. _The Atlantic_ magazine has created a searchable resource to find out if your works are included in the LibGen pirate database (I’m not aware of any similar resource for PiLiMi). The Authors Guild’s step-by-step guide on how to file a claim. SFWA’s Anthropic FAQ. The Authors Guild has a detailed explainer on the settlement. SFWA is conducting a survey for authors who have been impacted by the Anthropic downloads. You don’t need to be a SFWA member, or a speculative fiction writer, to fill it out. The only valid source of US copyright registration is the US Copyright Office. There are also many faux registration and timestamp services that purport to “register” your work—but they aren’t a substitute and have no legal standing, and some are scams. Here’s an example. ### Share this: * Click to share on X (Opens in new window) X * Click to share on Facebook (Opens in new window) Facebook * Click to share on Threads (Opens in new window) Threads * Click to share on Reddit (Opens in new window) Reddit * Click to share on Mastodon (Opens in new window) Mastodon * Click to share on LinkedIn (Opens in new window) LinkedIn * Click to share on Bluesky (Opens in new window) Bluesky * Click to share on Tumblr (Opens in new window) Tumblr * Click to email a link to a friend (Opens in new window) Email * Click to print (Opens in new window) Print * Click to share on Pinterest (Opens in new window) Pinterest * Click to share on WhatsApp (Opens in new window) WhatsApp * More * * Click to share on Telegram (Opens in new window) Telegram * Click to share on X (Opens in new window) X * ### Like this: Like Loading... ### _Related_ https://writerbeware.blog/2025/10/31/the-anthropic-class-action-settlement-what-you-need-to-know-right-now/