First AI copyright infringement settlements - both within a week: Anthropic and Eleven Labs avoid risking billion-dollar valuations.
- Josh Waterston
- 6 days ago
- 3 min read
Big news: Vacker v. Eleven Labs and Bartz v. Anthropic have settled. Why did they settle, and what does this mean for the other 46 AI copyright infringement cases currently pending?
Vacker v. Eleven Labs: Aug. 20th Stipulation to Stay Proceedings
Bartz v. Anthropic: Aug. 26th Order re: Settlement in Principle
First, there are a few common themes:
The alleged copyright infringement was blatant and intentional
The companies faced both reputational and financial harm
Both were raising money based on valuations of billions of dollars
Their losses could potentially damage or destroy the business model (some companies training on everything, regardless of copyright) of the generative AI industry itself – which copyright owners see as not necessarily a bad thing.
For Eleven Labs, the defendant allegedly copied the voices of two plaintiffs without consent, and removed technical protections and copyright management notices from audiobooks in order to train their models. According to the transcript of the March 16, 2025 hearing, Eleven Labs had received $280 million in funding, on a valuation of between $1 billion and $3 billion. In addition, it had already made news for the January 2024 President Biden robocall deepfake created using its platform. The risks apparently outweighed the cost of settling. Also, according to the stipulation: kudos to Jeff Kichaven Commercial Mediation for facilitating the settlement.
For Anthropic, things weren’t going well. The case was certified as a class action, and Judge Alsup denied Anthropic’s motion to stay pending appeal. Despite its June 2025 win on the issue of whether training its AI model on pirated books was fair use – a decision that made headlines – Judge Alsup detailed some pretty damning evidence on Anthropic’s willingness to blatantly pirate, scan, and retain books, which executives saw as fair use, but which Judge Alsup saw as copyright infringement. The plaintiffs were granted the right to add evidence to their argument against Anthropic’s attorney-client privilege defense, and a hearing was scheduled for today.
In his order denying the motion to stay, Judge Alsup wrote:
"Furthermore, Anthropic recopied only some “subsets” of the pirated works to use for training large language models, or LLMs, but kept them all. Which works in its collection were actually used to train LLMs, which were not, and why were all retained? Anthropic has refused to come clean on this, even now, and for all we know, most were never used (or not solely used) to train LLMs."
Bartz v. Anthropic PBC, No. 3:24-cv-05417, 2 (N.D. Cal. Aug 11, 2025)
All of this comes against the backdrop of Anthropic seeking to raise $5 billion on a valuation of $170 billion, after raising $3.5 billion on a $61.5 billion valuation in March 2025, according to CNBC’s July 29th article. This case was one of five against Anthropic in the Northern District of California.
Both Eleven Labs and Anthropic had plenty to lose by going to trial: the closure of their businesses and a potential dagger to the heart of their business models. From this perspective, settlement was the better option. Stay tuned to see how this affects the many other pending AI copyright infringement cases…
Do you agree with these outcomes? What do you think comes next? For questions about copyright, AI, licensing, or technology and IP law generally, feel free to get in touch with Wilftek attorneys Fred Wilf, Thomas Harper Kelly, Lewis Sorokin, or me.
Hat tip to Edward Lee and the fine folks at ChatGPTiseatingtheworld.com for staying on top of the latest developments, including Prof. Lee's deep dive into whether these settlements will be the first domino to fall. Keep up the good work.
