A US federal judge postpones Anthropic's $1.5B copyright settlement over concerns that class lawyers are striking a deal behind the scenes to force on authors
The federal judge overseeing AnthropicPBC's proposed $1.5 billion copyright settlement is concerned class lawyers are striking …
Bloomberg LawAnnelise Levy
Context & Ripple Effects
The proposed resolution followed an August filing that Anthropic had settled a class action tied to works in two pirate databases, and a subsequent filing described $1.5 billion plus interest for authors. The court’s intervention puts the mechanics of representing and compensating the class, not just the settlement amount, at the center of the case.
This procedural challenge was an early hurdle in a matter later described as the first major US case of its kind to settle. That makes judicial scrutiny of the class process consequential alongside the underlying copyright allegations.
First-order effects
The proposed settlement is postponed while the judge examines whether class counsel’s approach fairly represents authors, delaying a court-approved path to payment and release of claims.
Anthropic and class lawyers must address the court’s concerns before the agreement can move forward; authors retain an unresolved role in determining whether to participate or object.
Second-order effects
The delay raises the importance of notice, objections, and opt-out rights for authors, rather than treating the announced $1.5 billion figure as the case’s final outcome.
Other parties negotiating copyright class resolutions over AI training material may face closer attention to whether counsel can bind a dispersed group of rightsholders on acceptable terms.
Third-order effects
If courts continue to demand more scrutiny of representation in AI-related copyright classes, settlements may become less of a one-step mechanism for clearing large training-data disputes and more dependent on durable author-consent processes.
The case points to a broader split between resolving liability with a class settlement and securing legitimacy among individual creators; how judges handle opt-outs will shape how broadly such deals can settle future claims.
The trend:AI copyright disputes are moving from arguments over training data toward the practical question of whether class-action settlements can deliver credible, collective resolution for creators.
And please consider that the settlement needs to be approved by Judge Alsup, who is not happy about some of the details. news.bloomberglaw.com/ip-law/ anthr...
Judge Alsup said class members “get the shaft” in many class actions once the monetary relief is established & attorneys stop caring. He told the parties that “very good notice” must be given to class members. 📚⚖️👇 🏾 #anthropic news.bloomberglaw.com/ip-law/ anthr... …
Even the judge is like “nah this is a stitch up”. — Back to the drawing board, Anthropic and Author's Guild. Maybe have another swing that doesn't leave most authors in the dust. — news.bloomberglaw.com/ip-law/ anthr...
Here's the part where copyright law's private-law/property mechanics smack everyone in the face with the reality that it is not a general-purpose regulatory law, even with a large amount of money on the table [embedded post]
The judge evidently told the parties that “very good notice” must be given to “class members to ensure they have the opportunity to opt in or out, and protect Anthropic from potential claimants coming out of the woodwork later.” 3/
But that said, it doesn't appear that the judge is siding with the authors protesting the settlement. Instead, the judge “said class members ‘get the shaft’ in many class actions once the monetary relief is established and attorneys stop caring.” 2/
Wow!!! Major news!!! The judge in the Anthropic AI case is “concerned class lawyers are striking a deal behind the scenes that will be forced ‘down the throat of authors.’” — This is what myself and many other authors were saying about the settlement. 1/ — news.bloombergla…