DataPacts is an independent, two-sided reference on data acquired for AI training: how AI developers obtain training data and how data holders license or sell it. It is a reference, not a marketplace, and it does not negotiate, run or close any data sale. It also takes no cut of any sale, and no buyer or program can pay for a place on a page.
The site is the independent reference on AI training data licensing, written for the holder deciding whether to license and for the buyer, adviser or researcher who needs the same facts, with every statement tied to a primary source and a date.
Written by the DataPacts Editorial Team; every figure follows how DataPacts sources, dates and checks every figure.
What DataPacts covers
DataPacts covers every step between a data holder and an AI developer, from who sells and who buys to the law, the contract and the price. The reference is organized in eight areas:
- Seller types, from operating and defunct companies to publishers, creators and institutions
- Data types, from documents and code to audio, video and sensor data
- Buyers and their programs, with the date each program was checked
- The seller process, from readiness and rights audit to delivery and payment
- Law by jurisdiction, including copyright, privacy and insolvency rules
- Court cases about training data, with docket numbers and current status
- Licensing deals, stated only where a party has published the terms
- Market data and prices, shown as dated claims
Outside the scope are consumer data brokers for marketing, model-training tutorials and labeling-software reviews, because none of them concerns how training data changes hands. Page counts are not stated, as the reference grows with each batch of pages.
Who the reference is written for
The reference is written for both sides of a data license: the holders deciding whether to sell, and the buyers, advisers and researchers who need the same facts. Two-sided means one page that serves both perspectives, not two versions of the truth, so a seller and a buyer read the same terms, risks and dates. Four groups use it:
- Data holders: operating, defunct and insolvent companies, publishers, creators and institutions
- Advisers, liquidators, administrators and investors who handle or value data assets
- Buyers and researchers who compare programs, terms and legal exposure
- Journalists who need dated, primary-source facts
Readers who want to see the holders' side in detail can start with who sells data to AI companies and the options each type of seller has.
What independence means at DataPacts
Independence at DataPacts means that no buyer, broker or referral program can pay to be included, placed higher or described differently. The table below lists the seven commitments that follow from this rule (as of 2026-11-09) and the policy that sets each one.
| Commitment | What it means on the page | Policy source |
|---|---|---|
| No brokerage | DataPacts does not negotiate, run or close data sales | Editorial policy, Disclosure |
| No cut of any sale | No percentage or fee from any data sale between a holder and a buyer | Disclosure |
| Inclusion never depends on fees | Every buyer found is covered, whether or not it pays referral fees | Editorial policy |
| Neutral order | Lists run alphabetically or by a stated neutral criterion (date, data type), never by payout | Editorial policy |
| No endorsement labels | No buyer is called a partner, recommended, verified, trusted, top or preferred | Editorial policy |
| Buyer figures as claims | Payouts and fees are written as the buyer's own statement with a check date | Methodology |
| Not selling is an outcome | Decision pages list alternatives, including not selling or licensing at all | Editorial policy |
Because inclusion does not depend on fees, the directory of every buyer and program, compared without fees lists a program that pays no referral fee exactly as it lists one that does.
Decision pages follow the same rule on the holder's side. Each states benefits, risks and downsides, and names the alternatives, which include keeping the data, deleting it or declining to license it. The pages are information, not legal or tax advice. Where a decision depends on facts the reference cannot know, such as the rights a company holds in its own data, the page says so and points to the step that checks it.
Revenue sources and the no-cut rule
DataPacts is funded by company data enquiries, framed as requests for an independent introduction to suitable buyers, by an introducer form and by a newsletter. Later referral programs are used only when disclosed, and as of 2026-11-09 none has been joined and no page carries a referral code or tracking link.
DataPacts never takes a cut of any data sale, and the details are set out in how DataPacts is funded and what it never does.
Who writes DataPacts
DataPacts is written and maintained by the DataPacts Editorial Team, and no page claims to be written or reviewed by a named expert. The byline is the same on every page, so a reader never has to guess who stands behind a statement. No person, owner company or production detail is named, because the reference answers for its content as an editorial team and not through individual authority.
No page offers legal or tax advice, and none replaces a lawyer, an accountant or a tax adviser. A reader who plans a license should take advice for the jurisdiction and the data concerned.
Sourcing and dating, in brief
Every figure on DataPacts comes from a primary source and carries the date it was checked. Primary sources are laws, regulators, court dockets, filings, the buyers' own pages and interviews with named people, while Wikipedia and competitor sites are never a source. A buyer's payout or fee is written as that buyer's own claim, and the full rules are set out in how we source, date and check every figure.
Contacting the editorial team and related trust pages
The editorial team is reached at editorial@datapacts.com, for corrections, coverage questions and anything about how the site works. The related trust pages are: