Sell Workflow DataGet a quote

Who removes the names? De-identification by buyer

All four strip identities; how much they explain differs. Scale AI works on your side before extraction. micro1 and Labelbox do the work themselves and leave location unstated. Nyne promises de-identified data prior to release and names no one.

This sheet is independent: none of the companies on it is affiliated with us, and none has endorsed it.

Sheet last verified 7 November 2026, against every buyer's own page.

Who removes what, and where?

De-identification terms published by each workflow data buyer
BuyerWho de-identifiesWhere or whenYour reviewOwnership after
micro1micro1Not publishedSeller can inspect representative samples; synthetic rewording on some setsUnderlying data stays the seller’s
Scale AIScale AIOn the seller’s side, pre-extractionSeller signs off every packageNot stated as such; kept under seller governance
NyneNot publishedPrior to releaseSeller picks categories to hold backSet by the agreement
Labelbox PrismLabelbox, as part of Prism ingestionNot publishedFree scoping; first valuation in roughly two daysSeller retains control of data and rights

Condensed from each buyer's wording, opened 11/07/2026. Disclosure: a buyer here may later pay us a fee for an introduction we made; no cell turns on that. micro1 adds two limits of its own: customers must be labs or enterprises based in the US or allied nations, and no resale or distribution happens while identifying details remain.

Before export

What should you remove before any buyer sees a sample?

A first pass on your side, whatever the buyer does afterward. It is also the evidence that you took care, if a question ever comes up.

Fields to remove or tokenize before export
FieldWhat to do
Staff names, emails, phone numbersReplace with stable tokens (agent_07) so the sequence still reads
Customer names, emails, addresses, account numbersReplace with tokens; drop free-text signatures and footers
Payment detailsRemove entirely; no buyer needs them
Client names in project recordsReplace or exclude the project if the client contract forbids licensing
Attachments and screenshotsUsually excluded; they hide personal details that text filters miss
Credentials, API keys, internal URLsScan and strip before any export

Scrubbing names does not answer whether the records were yours to share. Customer personal details and client property remain off the table unless contract and statute both permit them; your counsel confirms that before export.

Sources behind this sheet (4)

Questions

Questions about de-identification

Which buyer de-identifies before data leaves the company?

Only Scale AI says so: anonymization runs on the seller’s side ahead of extraction, and nothing ships without sign-off per package. The rest leave location unstated.

If the buyer de-identifies, why do it ourselves?

Because you are the one warranting that you had the right to share what left your systems. Removing obvious personal details first narrows what a buyer’s process has to catch, and what could go wrong in transit.

Can we see the de-identified data before it is used?

With micro1, representative samples; with Scale AI, sign-off per package. Nyne and Prism print no review step, so write one into the agreement.

What is a synthetic rewrite?

A technique micro1 mentions for some datasets: rewording text so the process survives while the original phrasing does not. Ask which of your sets it would touch.

Get a quote

Takes about two minutes. Give a ZIP code, a reply address and a few facts about your records; the reply names each buyer whose printed criteria your company looks set to clear.

Describe the records; never attach or paste them. No data, exports, screenshots or samples belong in this form.

Plain disclosure: introductions are our business, and the company we introduce you to may pay us.

1/3

Where is the company based?
Get a quote