A $16.3B market, and no chain of title.
The scraping era ended because supply ran short, not because a court closed it.
What replaces it is licensed, non-public, and sold as rights-cleared — the least examined phrase in this market. Every seller in it is an intermediary, not a culture.
A license proves somebody signed. Clearance proves the signer had the right to. The AI training data market has been buying the first and calling it the second.
High-quality public text is projected to be effectively exhausted for training between 2026 and 2032. That constraint — not conscience — is what created a market for consent.
| Seller | Buyer | Reported value | What was actually sold |
|---|---|---|---|
| News Corp | OpenAI | $250M over 5 years | Five years of renewable access — training plus display, not a corpus handover |
| Google, OpenAI | $203M aggregate contract value at IPO | Other people's posts, training plus grounding | |
| Shutterstock | OpenAI, Meta, Google, Amazon, Apple | $104M AI licensing revenue in 2023 alone | Contributor images, video, audio, 3D |
| The New York Times | Amazon | $20–25M per year | Archive plus real-time answers |
| Financial Times | OpenAI | $5–10M per year | Archive plus attributed display |
| Museums, archives, community collections | — | $0.00 | Not registered. Not titled. Not offered. |
Reported figures per sources 1–2. Most contracts are private; treat any total as a floor. Final row is a MADE CX observation, not a reported transaction.
Warner, Universal and BMG converted music suits into licenses.3 Valid — and incomplete: in June the musicians' union sued both labels for licensing members' recordings without paying the performers.4
Every settlement terminates at the intermediary, the only party the contract can see.
A dataset is only as cleared as its weakest link in the chain of title — and most cultural chains were never written down.
Founder's Note No. 08On 17 August 2026 Round Hill Music sued Suno and Anthropic in the Northern District of California, seeking up to $150,000 per willfully infringed work — each case potentially approaching or exceeding $1 billion — and stated it does not intend to settle.
I want it to succeed. But look at the shape: statutory damages, retrospective, on works a plaintiff can name and own. Round Hill controls 31,237 works; each complaint attaches 500.5
The benchmark is set. Anthropic's authors' settlement — the largest in U.S. history — resolved to roughly $3,000 per book. Once, for material already ingested.6
$3,000 per book is not a price. It is a settlement of a breach.
It arrives once, after ingestion, for a work already inside the weights. It creates no title, establishes no rate, and generates no second payment when the model earns for the next decade. A meter would have. Per-work settlement figure per source 6; the characterization is a MADE CX assessment.
Round Hill can sue over the works it owns, not over what they are made of. Blues structure, gospel cadence, the grammar a model reaches for to sound authentic — no registration, no plaintiff, no price.
That is the actual asset: consumed at industrial scale, invisible to the enforcement in use.
$CULTURE is the dataset copyright cannot describe: not a work but a lineage — which tradition, from whom, through which generation.
Chain of transmission is what a buyer needs for indemnity and a model for attribution — and scraped data cannot supply it. A community holds it; an intermediary does not.
The Blackchain Creative Identifier titles the asset. A work, a form, a mark, or a lineage receives a persistent identifier and a chain of record.
A standardized, priced, metered permission instrument. Scoped by use class, term-limited, auditable, and revocable on breach.
Thirteen sovereign registries across six chambers hold the record. Registration is what converts a claim into a searchable, provable asset.
Custody, settlement, and distribution to the originator of record — so payment reaches the bottom of the chain, not just the top.
Museums and archives are being approached as image suppliers with a catalog to hand over. They are not image suppliers, and the market they are being invited into stopped working that way two years ago.
The repricing already happened. News Corp did not sell OpenAI an archive; it sold five years of renewable access.2 Corpus sales gave way to licensed feeds.
Disclosed archive deals clear in the millions to tens of millions per annum with no cultural-origin premium. Lineage-documented archives should clear above that band.
A Black cultural archive should never be sold as a corpus.
A corpus is delivered once and priced once. Whatever it earns for the next decade, it earns for somebody else. A feed is termed, metered, audited, and repriced at every renewal — and it is the only structure under which the archive is still yours the morning after the deal closes. The corpus-versus-feed framing is a MADE CX assessment; the deal structures behind it are reported at source 2.
| CUL class | What the buyer gets | Pricing basis | Reversibility |
|---|---|---|---|
| Class 01 · Train | Corpus ingested into model weights | Per registered unit, term-bounded, priced at ceiling | None once trained — price accordingly |
| Class 02 · Ground | Retrieval at inference time | Metered per call | Revocable |
| Class 03 · Display | Surfaced in output with attribution | Metered, plus enforceable credit obligation | Revocable |
| Class 04 · In-Style | Output generated in the registered form or genre | Royalty on output revenue, not a flat fee | Revocable |
CUL class structure and pricing bases are the MADE CX framework, not observed market rates. Comparable per-annum archive figures are reported at sources 1–2.
This is not either/or. Round Hill's suit is a legitimate and necessary enforcement of a right that exists. The point is what that mechanism structurally cannot do — and what has to exist before the next corpus is contracted.
| Litigation (the Round Hill model) | CPRS enforcement | |
|---|---|---|
| When it acts | After ingestion | Before ingestion |
| What you must prove | Ownership, access, copying, willfulness | That you registered |
| Coverage | Registered works only — 500 bellwether exhibits at a time | Works, forms, marks, and lineage |
| Timeline | Years to trial; outcome uncertain | At the point of license |
| Remedy | Statutory damages up to $150,000 per willful work, or settlement | Priced CUL, audit right, revocation on breach |
| Payment shape | One-time, retrospective | Recurring, metered, forward-looking |
| Who receives it | The rights holder of record — the intermediary | The originator of record, through custodian settlement |
| Cost to pursue | Trial risk, contingency, counsel | Registration and metering fees |
| What it leaves behind | A judgment | A title, a rate, and a record |
Litigation column reflects the claims and remedies described in source 5. CPRS column describes MADE CX instruments in operation today.
Litigation asks you to prove you were robbed. A registry asks only that you wrote down what was yours.
Founder's Note No. 08The corpus of the next decade is being contracted this quarter. Everything below is available today, under instruments that are already live.
It compensates the rights holder of record. That is often not the person who made the thing. The AFM suit exists precisely because a valid label license did not reach the musicians on the recordings, and the labels' response was that AI compensation falls outside the existing agreement.4
A settlement is a payment. A registry is a plumbing change.
Markets price unregistered things every day once someone builds the registry. Land was priced before title existed; title made it bankable. Domain names, carbon, spectrum — none had a price until a register, a class system, and a clearing mechanism existed.
CPRS is not a claim against copyright. It is a voluntary, contractual layer that sits alongside it and covers what copyright's subject matter was never built to hold.
Because they demonstrably already do. The entire licensing wave of 2023–2026 is buyers paying for material they had previously taken, at rates that ran into the hundreds of millions.2 What moved them was not ethics; it was litigation exposure, supply scarcity, and the operational value of clean provenance.
All three of those pressures point the same direction, and all three are getting stronger.
Litigation will recover some of what was taken. But a judgment is not a title, and a settlement is not a rate.
License, don't ingest — that argument is won. The next one is open, and it will not stay open long.
Yours in the interest of the race,
Thirteen sovereign registries across six chambers, where $CULTURE is titled, priced, and metered as financial-grade property. Operating today.
Reported figures are cited to their source. Where a figure is a MADE CX framework, computation, or assessment rather than a reported market number, it is labeled as such.
MADE CX publishes editorial research and commentary on cultural property and commerce. Nothing on this site is an offer to sell or a solicitation of an offer to buy any security or interest, and nothing here is investment, legal, accounting, or tax advice.
Designations shown with a dollar sign — for example $CULTURE — are editorial identifiers used to name the subject of an analysis. They are not securities, digital assets, tokens, funds, or tradeable instruments of any kind, and no such instrument is offered or available.
Names, marks, and images of individuals, estates, and organizations appear for purposes of news reporting, commentary, and analysis. Their appearance does not imply affiliation with, sponsorship by, or endorsement of MADE CX.
© 2026 MADE CX. MADE CX, CPRS, and Culture Exchange Standard are trademarks of MADE CX. All other marks belong to their respective owners. Corrections and rights inquiries: hi@madecx.info.