The outline
| Clause | What it says | What to push for |
|---|---|---|
| 1. Definitions | "Licensed Data", "Permitted Purpose", "Derivatives" | Narrow definitions; ownership never transfers by definition |
| 2. Grant of license | Rights the buyer receives | Non-exclusive, non-transferable, limited to the Permitted Purpose |
| 3. Scope schedule | Systems, teams, date ranges, file types | Attach as an exhibit you control |
| 4. Exclusions | What is never delivered | HR, health, privileged, union, payment card, client-owned data |
| 5. De-identification | Who removes personal data, to what standard | Buyer performs and certifies; named standard |
| 6. No re-identification | Ban on re-identifying people | Public commitment, flows to all downstream recipients |
| 7. Sample review | Your approval before use | Written approval right |
| 8. Security | Access, isolation, devices, breach notice | Named controls; short breach-notice window |
| 9. Onward transfer | Sharing with labs or third parties | Disclosure of recipients; same terms flow down |
| 10. Retention and deletion | How long, how deleted | Fixed term; written deletion certificate |
| 11. Fees and payment | Amount, milestones, timing | Upfront or milestone payments; clear triggers |
| 12. Warranties | Your promises about rights | "To our knowledge" qualifiers |
| 13. Indemnities and liability | Who covers claims | Mutual; capped at fees paid |
| 14. Term and termination | How it ends | Termination for breach with deletion |
| 15. Publicity | Use of your name | No use without written consent |
Go deeper: the 15-point agreement checklist.
Frequently asked questions
Is there a template for an AI data licensing agreement?
This page gives a clause-by-clause outline to brief your lawyer or review a buyer's draft. Have counsel draft or review the actual contract.
Should an AI data license be exclusive?
Generally non-exclusive unless you're paid a premium for exclusivity, so you can license the same data to other buyers.