Anthropic's $1.5 billion settlement in the Bartz v. Anthropic class action is moving toward disbursement, and the fight over who collects is already exposing fault lines between authors and their own publishers2. Meanwhile, the litigation pipeline keeps growing: two more newsrooms sued OpenAI and Microsoft on September 51,3, and nearly 400 local newspapers have also recently sued the two companies. ANALYSIS Together, these strands sketch the emerging, messy economics of AI training-data compensation.
Why it matters
Anthropic's $1.5 billion settlement is now putting specific dollar figures on per-work compensation: authors of nearly 500,000 titles will be paid $3,000 for each pirated work. How that money actually flows, and who intercepts it, will set expectations for every future negotiation between AI labs and rights holders. ◆ The author-publisher disputes now surfacing suggest that even a resolved lawsuit does not resolve the economics.
The big picture
Under the settlement's terms, authors of nearly 500,000 titles will be paid $3,000 for each pirated work. If a book is still in print with a traditional publisher, the payment is split 50-50 between author and publisher; if the book was self-published or rights reverted, the author receives the full amount. According to a plaintiffs' status report, the initial net payout will be $2,200 per work after deductions, with disbursements expected by November 15, 20265,6. That amount could increase later depending on subsequent claims, allocation disagreements, and two filed appeals.
But writers are already reporting that publishers appear to be claiming shares of payments they may not be entitled to. Mystery and thriller author April Henry posted: "WTF is HarperCollins playing at? They claimed one of my books on the Anthropic Settlement that reverted back at least 17 years ago AND on the same day I got a credit alert saying they had been added as my employer! (which they never were)". Victoria Strauss, writing at the blog Writers Beware, said she has been receiving author complaints falling into two broad categories, one involving publishers seeking payments on works whose rights have reverted.
ANALYSIS The gap between the $3,000 headline figure and the $2,200 net payout already reflects the friction of class-action mechanics. Publisher overclaims threaten to widen that gap further for individual authors, turning a settlement designed to compensate creators into another intermediary extraction point.
On the litigation side, The Seattle Times and Newsday filed suit against OpenAI and Microsoft on September 5, alleging the companies used their journalism as training data without permission and that AI products often reproduce passages from their reporting in response to user queries. The lawsuit described generative AI as "a snake eating its own tail" that could "destroy the very organizations" that produce the content it is trained on. Nearly 400 local newspapers have also recently sued OpenAI and Microsoft.
The Seattle Times case carries a particular wrinkle: Microsoft and OpenAI are funders of some of the newspaper's journalism projects and fellowships4. A Microsoft spokesperson told GeekWire that the company is "surprised by the lawsuit" but is "always happy to sit down and explore solutions to this type of dispute".
ANALYSIS Two dynamics are running in parallel and reinforcing each other. The Anthropic settlement established a judicial framework: a judge ruled that training AI models on copyrighted material is legal under fair use, but pirating that material is not. That distinction gives future plaintiffs a narrow but defined lane. The Seattle Times and Newsday complaint, which alleges unauthorized use and verbatim reproduction, fits squarely within it.
The author-publisher payment disputes reveal a second-order problem that no courtroom ruling addressed. Settlement funds flow through existing contractual relationships between authors and publishers, and those contracts were written long before AI training was a commercial activity. Publishers asserting claims on reverted titles suggests the infrastructure for distributing AI-related compensation is being stress-tested in real time.
The fact that The Seattle Times accepted funding from the very companies it is now suing underscores how entangled content producers and AI firms have become. Microsoft's response, that it is "surprised by the lawsuit", paired with its offer to "explore solutions," fits a pattern of treating financial partnerships with publishers as buffers against legal exposure.
What's next
The Bartz v. Anthropic disbursements are scheduled to begin by November 15, 2026, with two appeals still pending. How allocation disputes between authors and publishers are resolved in the intervening weeks will shape whether the settlement functions as a template or a cautionary tale. The Seattle Times and Newsday suit adds to a docket that already includes The New York Times, Ziff Davis, Merriam-Webster, and Encyclopedia Britannica as plaintiffs against OpenAI. Nearly 400 local newspapers are also pursuing claims against OpenAI and Microsoft. ANALYSIS The settlement and the new suits together confirm that the cost of training data is now a line item; the unresolved variable is how much of each dollar reaches the creator rather than an intermediary.