Most creators who had work in the Anthropic pirated dataset could not claim a penny of the $1.5 billion settlement — not because their copyright was invalid, but because they could not prove ownership to the standard the process required. That gap — between having a right and being able to prove it — is the most consequential legal fact in modern IP.
You insure your car. You insure your home, your health, the things you own. Now think about the work that actually pays for all of it — the manuscript, the song catalogue, the photographs, the codebase, the designs. For most people who create for a living, that work is worth more than the car and the house combined, because it is the asset that earns the income. And it is almost always the one thing they protect with nothing at all.
In September 2025, Anthropic reached a proposed $1.5 billion settlement — still awaiting final court approval as this post publishes — that showed exactly what that lack of protection costs.
Read the settlement closely and a hard truth emerges: the people who could claim were, almost without exception, the people who could prove they owned their work. Everyone else — most working creators — was a bystander to the largest proposed copyright recovery in US history.
That dividing line is the most important line in modern IP. It is also the line Provlyn exists to put you on the right side of. Provlyn gives any creator independent, verifiable proof of what they created and when — the same kind of evidence that separated the claimants from the bystanders in the Anthropic case.
This post explains what the settlement actually decided, why most creators were excluded by the structure of it, and how prior proof of ownership closes that gap — permanently, and for a fraction of what you already spend protecting things worth far less.
Anthropic won on AI training; it lost on how it sourced its training data. Judge William Alsup ruled in June 2025 that training Claude on lawfully acquired books was fair use. The infringement lay in obtaining roughly 500,000 books from pirate libraries — Library Genesis and Pirate Library Mirror — not in what those books were used for.
Anthropic faced a December 2025 trial on the piracy question and the statutory damages that could follow. With exposure potentially reaching into the hundreds of billions, the company settled. The result was approximately $3,100 per work, across around 482,000 books. The settlement is not a verdict on whether AI training is lawful. It is a verdict on unlawful sourcing — and a demonstration of what proven ownership is worth when that sourcing is challenged.
The certified class covered rightsholders whose books were in the pirated libraries and were registered with the US Copyright Office within the window required to access statutory damages. In practice, 92.77% of the roughly 482,000 eligible works were claimed — because they were books with ISBNs, moving through a publishing infrastructure that generates registration as a matter of routine.
Now read that requirement the other way. The class included people with formal proof of ownership. It excluded, in any practical sense, everyone without it. A photographer, illustrator, independent musician, software developer, course creator, or self-published writer whose work sat in the same pirate libraries — but who had no Copyright Office registration, no ISBN — was not a class member. Their copyright was no less valid. Their work was taken in the same way. But they had no practical route into the recovery, because they could not produce the one thing the process ran on: independent proof of what they created and when.
The difference prior proof makes — Anthropic settlement as a case study
With formal proof of ownership | Without formal proof of ownership | |
Work in the pirated dataset? | Class member. Eligible to claim ~$3,100 per work. | Not a class member. No route into the recovery. |
Copyright valid? | Yes | Yes — but unprovable to the required standard |
Work taken in the same way? | Yes | Yes |
Practical outcome | Claimant | Bystander |
Across every major AI copyright case, the pattern is the same: creators who can prove what they made, when they made it, and that it pre-existed the alleged use are the ones who recover, settle, and license from a position of strength. The creators who cannot are spectators to a process built entirely on evidence they do not hold.
One thing needs to be precise here, because it protects you from a false hope. In the United States, accessing statutory damages requires registration with the US Copyright Office within a set window. No third-party timestamp substitutes for that registration. Prior proof of ownership does something different and broader: it establishes, independently and verifiably, that a specific work existed in a specific form on a specific date, created by you. That is the foundation every ownership claim rests on — and in the majority of disputes a creator actually faces, outside the narrow question of US statutory damages, it is the operative evidence.
Provlyn gives any creator the evidential foundation that formal publishing gives a book — without a publisher, a registration process, or any administrative overhead. When you deposit a file, Provlyn generates an independent, verifiable record of exactly what that file contained and when it existed. That record cannot be altered after the fact, because it is secured across three independent systems simultaneously.
What Provlyn does at deposit — in order:
Place that against the Anthropic fact pattern. A creator holding a Provlyn certificate is no longer reduced to asserting authorship and hoping to be believed. They hold an independent, verifiable record of exactly what they created and when. If their work surfaces in a training dataset, a leak, or a competing product, they can prove it — to the evidentiary standard the entire process turns on. They have moved from bystander to claimant.
Prior proof of ownership is the piece underneath everything else — the documentation that makes a claim, a settlement, or an insurance pursuit possible in the first place. Specialist IP insurance, which exists and is widely sold, typically starts around £1,000 to £1,200 a year for a UK business — and even if it funds the legal fight, it assumes you can already prove what you own. Prior proof is what that assumption rests on, and it costs a fraction of any of those premiums.
To be clear, Provlyn is not insurance and does not pay out. It is the proof your case depends on — the evidence an insurer, a court, or a counterparty assumes you already have. Think of it the way you think about keeping receipts for the valuable things you own: not the cover itself, but the thing that makes the cover worth anything.
The Anthropic settlement established a number — around $3,100 a work — but more importantly it established a principle. AI systems that train on pirated work face damages at scale, and the money flows to the rightsholders who can prove ownership. The case is, in essence, a live demonstration of who collects when the question is finally asked in court.
A deposit made today establishes your timeline today, with mathematical certainty. The same deposit made after a dispute begins is worth far less, because its timing is now in question. The creators best placed for whatever comes next are the ones who established their record before they needed it. The question is whether you have established yours.
Establish prior proof of ownership for your creative work today
Provlyn provides cryptographic prior proof of ownership for any digital artefact. Each deposit is hashed with SHA-256, timestamped by an accredited Trust Service Provider under RFC 3161, qualified under eIDAS Article 41, and anchored on the Bitcoin blockchain. The vault certificate carries a legal presumption of accuracy under the EU eIDAS Regulation, is admissible evidence in UK and EU jurisdictions, and remains valid permanently, independently of Provlyn's continued operation. Start your free 7-day trial at www.provlyn.com. No credit card required.
Blockchain-anchored timestamps and court-admissible certificates. Prove it. Permanently.
Get started free