Data licensing guide

What happens to your data after you license it?

Quick answer

Your data is usually de-identified, packaged and passed to AI developers for research and training, under your agreement. Your own records don't change. micro1 says data is retained only for agreed periods and deleted on schedule, and Mode says AI developers don't receive unfiltered records. Deleting a dataset doesn't reverse what a trained model has already learned, so treat the licence as lasting in practice. Pin down retention, onward transfer and proof of deletion in the contract.

Originals
stay with you, unchanged
4
stages after signing
Can't un-train
deletion doesn't reverse a trained model
In writing
get retention and deletion terms

Most guides stop at the payment. The more useful question for a careful owner is what happens next, and which parts can't be undone.

01The four stages

  1. Preparation. The agreed copy is filtered and de-identified. Who does this, and to what standard, differs by programme.
  2. Delivery. The dataset goes to the AI developers the programme works with, under whatever your agreement allows.
  3. Training. Developers use it to train or evaluate models. This is the step you can't reverse.
  4. Retention and deletion. Copies are kept for an agreed time, then deleted, if your contract says so.

02What the programmes say

ProgrammeWhat it publishes
micro1Data is retained only for agreed periods and deleted on schedule
ModeSensitive material is removed, identifiers removed or transformed, AI developers don't receive unfiltered records, and it doesn't publish source records
MercorAutomated anonymisation, and no lab sees raw data

Each programme's own pages, checked 30–31 October 2026. Your contract, not the website, is what binds them.

03What doesn't change

Your originals stay in your systems, and your business carries on as before. Mode says no operational change is needed. You can usually license other data later, unless you've agreed exclusivity.

04What you can't undo

A deleted dataset doesn't un-train a model. Once a model has learned from records, removing the source files doesn't take that learning out. So think of a deletion clause as covering copies of the data, not what's already been learned. Decide what to include on that basis.

That's the strongest reason to get the scope right before anything moves.

05Five things to pin down in the contract

  • Retention: how long each party keeps copies.
  • Deletion: what's deleted, when, and how it's confirmed to you.
  • Onward transfer: who can receive it, and whether they can pass it on.
  • Trained models: what can be done with them after the licence ends.
  • Incidents: who tells you about a breach, and how quickly.

06If your situation changes

  • You sell your company: a buyer will want to see this agreement. See before selling your business.
  • The buyer is acquired or fails: ask what happens to your data, and whether the licence can be transferred.
  • A person asks to be removed: if any personal data was in scope, you need a way to respond. See UK GDPR.

This isn't legal advice.

07Questions people ask

Can I get my data back?

You never lose your originals. Copies can be deleted under the contract, but a model trained on them won't un-learn what it learned.

Who ends up with my data?

It depends on the programme and contract. Ask who receives it, in what form, and whether it can be passed on again.

Could an AI model repeat my company's content?

Ask the programme what it does to reduce that risk. De-identification and filtering help, but get the answer for your data in writing rather than assuming.

Sources

Our links to these programmes carry our referral code, so the programme can see the introduction came from us. The programme may pay us a referral fee, and you pay nothing.

Check which programmes you’re likely to matchA few short questions, about a minute. Free, no sign-up. The programme decides who qualifies.trackimpact.io/data →