Find a buyer

From published deals and terms

AI data licensing: how the deals and terms work

AI data licensing lets a company or creator rent data to an AI developer under set terms instead of selling it outright. Here is how the published deals handle exclusivity, length, payment, permitted use and consent.

Lawyer and company executive reviewing a printed licensing agreement

The key terms

Six terms every AI data licensing deal sets

Each term below comes with a real example. Published AI data licensing deals vary a lot, so treat these as reference points, not a template.

Exclusivity

Training licenses are usually non-exclusive. Calliope Networks says its training licenses are non-exclusive and exclude exhibition rights. Informa's Microsoft deal is non-exclusive too, so the same content can be licensed again.

Term length

Reddit's January 2024 deals run two to three years. Informa's Microsoft deal runs from 2024 to 2027, and Shutterstock's OpenAI deal runs six years. AWS Data Exchange subscriptions run from 1 to 36 months.

Payment structure

Deals mix upfront and recurring money. Informa took a $10 million-plus initial fee plus recurring payments through 2027. IAC described a fixed plus variable payment from OpenAI. Snowflake listings can charge per query, monthly or by subscription.

Delivery

Platform data often flows as ongoing access, not a single file. Reddit's S-1 describes continuous Data API access plus quarterly data transfers over the contract.

Training or display

Some licenses cover training, others cover showing content in AI answers. Getty's Perplexity deal covers displaying images with credit and links. The FT and Dotdash Meredith deals cover attributed content in ChatGPT and model improvement. Created by Humans sells training, reference and transformative rights separately.

Use limits and attribution

Informa's agreement limits verbatim text extracts and stresses citation. The RSL standard lets publishers post machine-readable terms covering attribution, pay-per-crawl and pay-per-inference.

Pooled content

Consent and revenue splits

Business data needs its own guardrails. micro1's data partnerships agree on the approved data scope, security standards, anonymization and internal approvals before a company shares anything.

When a platform licenses content from many people, it has to get consent and split the money. The Dataset Providers Alliance calls for opt-in consent for using data and likeness in training. HarperCollins' Microsoft offer required author opt-in.

Splits are usually pro rata. Shutterstock pays contributors a 20% average royalty on data-licensing revenue, shared by how much of their work is in each dataset. Wirestock splits each dataset payment 50/50, with the creators' half divided by item count.

Opendatabay charges commission on self-serve sales, pays a fixed provider price on brokered deals, and still charges commission on deals it introduced that close off the platform. Read that clause before you sign up anywhere.

Best for: Companies with operational data: SOPs, CRM records, project histories

  • Pays for operational data
  • Scope and redaction agreed first
  • $500M gross run rate (reported)

micro1 is a data lab that makes expert human data, RL environments and evaluations for frontier AI labs. Its data partnerships program licenses companies' operational data, such as SOPs, knowledge bases, CRM data, project histories and QA processes. The data scope, anonymization and internal approvals are agreed before anything is shared. TechCrunch reported in August 2026 that micro1 reached a $500 million gross run rate.

Checklist for a data licensing agreement

  • Exactly which data, in what format, delivered how and how often.
  • Exclusive or non-exclusive, and in which fields.
  • Permitted uses: training, retrieval and display, or both.
  • Term, renewal and what happens to copies when it ends.
  • Upfront fee, recurring fee or usage-based payment.
  • Attribution and limits on verbatim output.
  • Who handles personal data and consent.

Where AI data licensing happens

Large publishers and platforms sign directly with AI labs. Everyone else usually goes through a marketplace or broker.

Compare the AI data marketplace options, read what AI data licensing deals have paid, or work out a price with our data valuation guide.

This page summarizes published deals. It isn't legal advice.

Questions

Common questions

Is AI data licensing usually exclusive?

Usually not. Calliope Networks states its AI training licenses are non-exclusive, and Informa's deal with Microsoft is non-exclusive too. Exclusivity is possible but limits who else you can sell to, so it should cost the buyer more.

How long do AI data licenses last?

Published examples run two to six years. Reddit's January 2024 deals run two to three years, Informa's Microsoft deal runs 2024 to 2027, and Shutterstock's OpenAI deal is six years. AWS Data Exchange subscriptions run 1 to 36 months.

What should a data licensing agreement include?

At minimum: what data and how it's delivered, whether the license is exclusive, permitted uses (training, display, both), term and renewal, payment structure, attribution, limits on verbatim output, and what happens to the data when the license ends. Have a lawyer review it.