# How to choose an AI data broker

> How to choose an AI data broker: eight checks on entity, terms, numbers, buyers, fees, data handling and contract, with dated examples from public buyer pages.

By Matt Burns, Vaultfish. Canonical: https://vaultfish.com/guides/how-to-choose-an-ai-data-broker  
Published: 2026-08-28  
Updated: 2026-10-14

Pick the buyer that lets you check the most. Seven checks cover most of it before you send a single record, and each takes minutes. In order:

1. A named legal entity
2. Live terms and privacy pages
3. Numbers that agree with each other
4. Named buyers
5. A published fee
6. A clear rule for when your data moves
7. A contract that sets liability

Each check below uses examples from public buyer pages.

Nobody has outside proof of what any buyer paid a business for its data. So these checks test whether a firm's story hangs together and how much it shows.

You make the call on each firm. The firms report their own figures, and none is a promise of what you would receive.

> **Data licensing means a company gets to use your records for a while, under rules. You keep ownership.** Think of renting a room. Before you hand over a key, you check who the landlord is, what the lease says and what it costs.

Figure: steps Check the entity -> Test the numbers -> Read the contract. Run all three before you sign anything.

People search for "data broker" when they mean an AI data company. Legally, a data broker resells consumer data and must register in some states. This site calls AI data companies buyers.

## Check 1: does the buyer name a legal entity?

Look at the footer and the terms. A named company gives you someone to hold to an agreement.

[SimpleClosure](/buyers/simpleclosure)'s terms name "Simple Closure Inc." Some referral sites give only a trading name, or an operator with no address or registration.

[Troveo](/buyers/troveo)'s business-data page footer shows "Troveo 2026" and no legal entity. Ask any of them for the contracting entity before a call.

## Check 2: are the terms and privacy pages live?

Open both pages and check the date. SimpleClosure's terms cover its Asset Hub. Troveo's `/terms` and `/privacy` paths returned 404. A missing page can be a site fault. Ask for the supplier agreement in writing.

## Check 3: does the headline match the calculator?

Most buyers show an estimator. Test its top end against the headline.

[Handshake AI](/buyers/handshake-ai)'s calculator example (100 employees, 10 years) gives $383,083 to $896,969. Its headline says $100K to $4M+. It calls the example "indicative only."

When a calculator tops out far below the headline, ask which companies reach the headline.

## Check 4: do the buyer's own pages agree?

Compare totals across its pages. Troveo shows $20M+ paid on its business-data page. Its content-owners page shows "$50M+" with no date.

Handshake AI shows $250M paid to partners on Handshake AI's [company referral page](https://joinhandshake.com/ai/data-partnerships/referral) and $500M+ in payouts on its fellowship page. These are company-wide totals, and none of them predicts what your company would receive.

Each pair may cover different groups. The pages don't say. Ask which group each number counts.

## Check 5: does it name who gets the data?

Marketplaces describe their buyers as a "frontier AI lab" or a "selected partner." Troveo's page uses "Mag 7 company." The Spirit Airlines bids from Google, [Mercor](/buyers/mercor) and [micro1](/buyers/micro1) were public only because a bankruptcy court ran the auction ([Time](https://time.com/article/2026/08/25/google-spirit-airlines-ai-data-RL/)).

Ask which kinds of buyer will see your data, and under what contract.

A micro1 referral partner made two points in a micro1 video (about 17:00 and 22:31). Owners are right to check that the company is reputable and that other companies are signing.

And the first question is who micro1 is. Ask who else is signing, and which lab ends up with the data.

## Check 6: are the fees and splits published?

Troveo shares revenue "on every sale" and charges companies no fees. It doesn't publish its split. An advisor paid on success should state its percentage before you sign.

SimpleClosure's terms point to fees in a separate order form, and its AssetHub page says it takes an unpublished percentage on a completed deal.

micro1 and Handshake AI publish tiers or ranges for what they pay. That gives you a figure to test. But a price range doesn't show what a middleman keeps. Get the percentage in writing.

## Check 7: when does your data move, and how is it prepared?

The companies differ here.

- [Troveo](https://www.troveo.ai/business-data) says it needs no OAuth access to your systems and no platform to install.
- Mercor connects to your tools and pulls the data through connectors.

micro1 staff said in an undated recorded forum talk that companies can't scrub first and micro1 must change the data itself. micro1's CFO said its engineers help your IT people with the transfer.

Mercor says no lab buyer or vendor sees raw data. Its own March 2026 incident update says a supply chain attack on LiteLLM affected it.

Whatever the answer, ask who holds the data, for how long and where it is deleted.

## Check 8: what does the contract say?

Daniel Goldberg, a partner at Frankfurt Kurnit, advises that a purchase agreement should "restrict how the buyer can use the data, allocate privacy risk, and address what happens if something goes wrong" ([Goldberg](https://technologylaw.fkks.com/post/102o455/selling-company-data-for-ai-training-what-to-know-before-you-say-yes)).

Ask about four things: the promises you must give, who pays if the buyer is breached, deletion on request, and exclusivity. Epoch's interviews put exclusive RL-environment deals at roughly 4x to 5x the non-exclusive price. Exclusivity changes value ([Epoch](https://epoch.ai/gradient-updates/state-of-rl-envs)).

## What about domain age and reviews?

Domain age is a weak signal. A domain registered this year leaves no track record to read. Mercor's domain, though, predates the company by two decades. Treat age as a cue to ask for references.

Reviews mostly describe worker platforms. We found none from companies that licensed business data. Ask each buyer whether a past licensor will talk to you.

A micro1 referral partner said in a micro1 video that she connects hesitant owners with owners who finished (about 19:34). Ask a reference four things:

- Who ran the export?
- How long did it take?
- What was removed?
- When did payment arrive?

_Domain age is a weak clue. A brand-new site can be honest, and an old one can still say "TBD" about fees._

## What should you do next?

Run the [rights check](/rights-check) and read the [risk map](/risks). Then bring the answers to a first call. The [buyer directory](/buyers) and [best AI data brokers](/guides/best-ai-data-brokers) list each firm's public terms. [AI data broker red flags](/answers/ai-data-broker-red-flags) ranks the patterns above.

For help getting your records and rights ready before any buyer sees them, see [managed readiness](/managed-readiness). We work from descriptions and counts, never your records.

Talk to a privacy or commercial lawyer in your state before you sign.

micro1 may pay us a referral fee if a company we introduce goes ahead with it. [How we are paid](https://vaultfish.com/disclosure)

[Apply to micro1's data partnership (paid link)](https://req.micro1.ai/post/74520e8b-4673-4da5-b855-53609aaf6710?referralCode=f82e519e-5d64-4e0e-9574-760c84371cf2&utm_source=referral&utm_medium=share&utm_campaign=data_referral) · [micro1's own data-partnerships page](https://www.micro1.ai/data-partnerships)

_Paid link. [How we are paid](https://vaultfish.com/disclosure)_

_Vaultfish helps you run the process. We are not lawyers and this is not legal advice. If you need counsel who knows this area, we can point you to one._

## Sources

- [company referral page](https://joinhandshake.com/ai/data-partnerships/referral)
- [Time](https://time.com/article/2026/08/25/google-spirit-airlines-ai-data-RL/)
- [Troveo](https://www.troveo.ai/business-data)
- [Goldberg](https://technologylaw.fkks.com/post/102o455/selling-company-data-for-ai-training-what-to-know-before-you-say-yes)
- [Epoch](https://epoch.ai/gradient-updates/state-of-rl-envs)