Data license
By Gareth HoyleLast updated How we research
A data license is a contract in which the owner of a dataset permits another party to use it for defined purposes while keeping ownership of it.
A data license is a contract that gives a buyer permission to use a dataset on stated terms while the seller keeps ownership of it. In AI deals the license sets what the buyer may do (train, fine-tune or evaluate models), for how long, in which territories and whether the rights are exclusive.
Under US copyright law an exclusive license is a transfer of copyright ownership and a nonexclusive license is not, according to 17 USC 101. Only a transfer needs a signed writing under 17 USC 204(a).
What does an AI data license need to cover?
An AI data license needs to cover more than a standard data agreement, because a trained model cannot easily forget what it learned. Nixon Peabody's October 2026 alert on buying AI training data lists the points buyers test:
- What is in the dataset
- Whether the seller has the right to transfer it for AI use
- Whether sensitive material can be removed without destroying its value
- What the buyer may do with derived assets such as annotations, embeddings, fine-tuned models and weights
- Who bears the risk through warranties, indemnities, caps and holdbacks
Our licensing agreement checklist walks through each clause, and AI data licensing explained compares a license with an outright sale or a revenue share. Take independent legal advice before you sign.