Skip to content
Sell My Data To AI
Menu

AI data license term sheet template

A one-document summary of the commercial terms for an AI data license, so both sides agree the main points before lawyers draft the full agreement.

In short

This term sheet template sets out the main terms of an AI data license before the full agreement is drafted. It covers the licensed data by reference to a manifest, license type and exclusivity scope, term, permitted and prohibited uses, derived assets, price in US dollars, payment timing, acceptance tests, de-identification, deletion, liability caps and termination. It is non-binding except for confidentiality, any exclusive negotiation period, costs and expiry.

Key takeaways

  • Defining the licensed data by an attached manifest, version and file hash prevents later disputes about what was promised.
  • The template treats exclusivity as its own term with a defined scope, and lets exclusivity end if payment is late.
  • Payment and acceptance run to fixed deadlines, so payment cannot be held open by delaying acceptance.
  • Only the confidentiality, exclusive negotiation, costs and expiry terms are binding. Everything else waits for a signed license.

A starting point for discussion, not legal advice

This template is written in plain English for US use and makes no claim about how a court would treat it. We are a broker, not a law firm. We tell you what buyers will check and help you prepare, but you should take independent legal advice before signing any license.

The files are built in your browser. Every detail to fill in is shown in [brackets].

How to use this template

  1. 1

    Fill in every [bracketed] placeholder, choosing one option where several are offered.

  2. 2

    Attach the data manifest as Schedule 1 and the exclusions as Schedule 2.

  3. 3

    Ask each buyer to respond against the same rows so offers can be compared.

  4. 4

    Have your own lawyer review it before either party signs.

The template

Data license term sheet

This term sheet summarizes the main proposed terms for a license of data from [Licensor legal name] ("Licensor") to [Licensee legal name] ("Licensee"). Except for the terms marked Binding, it is not a binding agreement, and neither party is obliged to enter into a license until both parties sign a definitive license agreement.

Date: [Date]

Parties and licensed data

TermProposed position
Licensor[Legal name, state of organization and address]
Licensee[Legal name, state of organization and address]
Signatories[Name and title of the person signing for each party]
Licensed dataThe dataset described in data manifest [Manifest name], version [1.0], dated [Date], with SHA-256 hash [Hash], attached as Schedule 1
ExclusionsThe categories listed in Schedule 2, such as [HR and payroll files, privileged legal advice, named client matters]
Future records[Not included] or [Records created up to (date)] or [Offered later as separate tranches]

License scope

TermProposed position
License type[Exclusive] or [Non-exclusive]
Scope of exclusivity[All AI uses] or [Field of use: describe] or [Model type: describe] or [Named competitors listed in Schedule 3]
Term[Number] years from the effective date of the license, or [perpetual]
Licensor's own useLicensor keeps the right to use the data in its own business, including for its own AI tools
Permitted uses[Pre-training, fine-tuning or evaluation] of [named models or model families]. Commercial deployment of the resulting models: [Yes or No]
Prohibited usesAny attempt to identify a person or business in the data, resale or sublicensing of the raw data, [use to build a product that competes with Licensor] and [other uses]
Derived assetsLicensee owns trained models and weights. [Cleaned, annotated or embedded copies of the data and evaluation sets built from it] may be used only within the permitted uses and may not be licensed to others
Onward transferOnly to [named affiliates] and to subprocessors bound by terms at least as strict as the license. No transfer to model partners or research partners without Licensor's written consent

Price and payment

TermProposed position
License feeUS$[Amount]
Payment timing[Upfront on signing] or [On acceptance] or [In milestones as below] or [Revenue share as below]
Milestones[Each milestone, its trigger and the amount due]
Revenue share[Percentage] of [defined revenue], with a minimum annual payment of US$[Amount], statements every [quarter] and a right to audit once a year
Payment deadlineWithin [30] days of [acceptance or invoice]
Late paymentInterest at [rate] a month, and Licensor may suspend delivery. Exclusivity ends if any payment is more than [60] days late
TaxesFees exclude [sales and use taxes]. Each party is responsible for its own income taxes

Delivery and acceptance

TermProposed position
Delivery methodA fixed export delivered by [secure transfer method]. No access to Licensor's live systems
Delivery date[Date]
Acceptance testsRecord counts, date ranges, formats and file hashes match the manifest within [tolerance]
Acceptance period[30] days from delivery, after which the data is treated as accepted

Personal data, security and deletion

TermProposed position
De-identification standard[Describe the standard, such as the CCPA definition of deidentified information or the HIPAA Safe Harbor method]
De-identification provider[Provider name], selected by [Licensor] or [both parties], with the protocol approved by Licensor
De-identification costPaid by [Licensor] or [Licensee] or [shared, with the split]
CertificationThe provider certifies each tranche before transfer
Re-identificationLicensee will not attempt to identify any person or business in the data and will bind every recipient to the same ban
Storage and access[Storage location], with access limited to [named teams]
Retention and deletionRaw copies deleted within [period] after [training or acceptance], with written confirmation
Security incidentsWritten notice within [72] hours of discovery

Risk allocation

TermProposed position
WarrantiesLicensor warrants its authority to sign and, to its knowledge, its right to license the data, subject to Schedule 2. Licensor warrants that the agreed de-identification method was applied
Indemnities[Scope of each party's indemnity], subject to the cap below
Liability capEach party's total liability is capped at [the fees received] or [an agreed multiple]. Model retraining costs are [excluded] or [capped at US$(amount)]
Holdback[None] or [US$(amount), released on (date)]

Publicity, termination and disputes

TermProposed position
PublicityNeither party names the other publicly without written consent. Where the law requires Licensee to disclose its training data sources, Licensor approves the description in advance where legally possible
TerminationOn termination, Licensee stops using the raw data and certifies deletion. Trained models survive [unless agreed otherwise]. Confidentiality, the re-identification ban, security obligations and payment terms survive
Governing lawThe laws of the [State of Delaware]
Disputes[The state and federal courts in (County, State)] or [Arbitration under (named rules) in (city)]

Binding terms

TermProposed position
Confidentiality (Binding)This term sheet and the negotiations are confidential and governed by the mutual nondisclosure agreement between the parties dated [Date]
Exclusive negotiation (Binding if included)[None] or [For (30) days from signing this term sheet, Licensor will not negotiate a license of the same data with anyone else]
Costs (Binding)Each party pays its own costs of negotiating and documenting the license
Expiry (Binding)This term sheet lapses if the parties have not signed a license by [Date]

Signatures

Acknowledged by the parties. Not binding except for the terms marked Binding.

Licensor
Signature:[Signature]
Name:[Name]
Title:[Title]
Date:[Date]
Licensee
Signature:[Signature]
Name:[Name]
Title:[Title]
Date:[Date]

Frequently asked questions

Is a term sheet legally binding?
This template states that it is not binding except for the sections marked binding: confidentiality, any exclusive negotiation period, costs and expiry. Whether a particular term sheet binds the parties can depend on its wording and how the parties act, so have your own lawyer review it before you sign. The full terms belong in the definitive license.
Why use a term sheet before the license?
A term sheet gets the commercial points agreed, such as price, exclusivity, term, payment and who pays for de-identification, before lawyers spend time drafting. It also lets you compare offers from different buyers on the same terms. Our licensing agreement checklist explains each clause the license should then include.
What should the exclusivity term say?
The exclusivity term should name what is exclusive (all AI uses, a field of use, a model type or named competitors), for how long, and whether you keep the right to use your own data, including for your own AI tools. Our guide to exclusive vs non-exclusive data licenses sets out the options and how buyers price them.
Can I compare offers with this template?
You can. Ask each buyer to state its offer against the same rows of this term sheet, so you compare the same records, uses, exclusivity, term and payment triggers rather than headline prices. The offer checker then shows how each amount compares with the published range for your industry.

Related guides

For data owners

Want a second pair of eyes?

A free valuation gives you a view of likely buyers, a realistic range and the rights issues to clear before a sale. Sellers pay nothing to be assessed.

Get a free valuationBook a free consultation