Guides · Published
Best AI training data licensing (2026)
Best AI training data licensing options in 2026, compared on ownership, cost, control and payout. Includes Troveo, Depositphotos, Smelted and more.
The best AI training data licensing options in 2026 depend on what you have to license. For internal business records such as support tickets, CRM history and project files, Smelted, Opendatabay and Troveo are the most relevant picks. For images, video and expert work, Depositphotos, Shutterstock, Troveo, Mercor and micro1 are better fits.
We make Smelted, so it's on this list; we judged it on the same criteria as the others.
How we picked
We compared each option on six criteria that matter to a company deciding whether to license data:
- What it licenses. Internal business records, media files, or expert work. A service built for one kind of data rarely suits another.
- Cost to the data owner. Whether the owner pays to take part, and what the sources say about payment.
- Ownership and license terms. Whether the owner keeps the data and grants a license, or hands over rights.
- Control. Whether the owner chooses what is included and who receives it.
- Privacy handling. Whether names and identifiers are removed before anything is shared.
- Payout evidence. Whether the option shows real figures or timelines that a buyer can check.
Where a provider's public material does not cover one of these points, we say so rather than guess.
The list
1. Smelted
Smelted is a free matching service that connects companies with AI data buyers for anonymized internal business records, such as support tickets, CRM history, SOPs, project files, code, CAD and lab data.
Best for: US companies with 50 to 500 employees and years of records in tools like Zendesk, Jira/GitHub, Salesforce/HubSpot, Bullhorn, AutoCAD/Revit or Benchling, in staffing, engineering, B2B software or science labs.
Strengths: The company pays nothing. The company keeps ownership, granting a license to a copy, and chooses what is included and which buyers it meets. Names, contact details and IDs are removed before anything is shared. Smelted finds matching buyers, gives a ballpark range, and introduces the company to two or three of them.
What to watch for: The value depends on headcount, years of records and how connected the data is, and no figure is given until a buyer has reviewed a sample. Payment usually comes a few weeks to two months after signing, so plan for one to three months from first call to payment. Buyers who take smaller companies pay less, so companies under 30 employees should expect lower offers. Smelted does not give legal or compliance approval, so check terms with your own counsel.
2. Troveo
Troveo runs a marketplace that licenses content from rights holders for AI training, across video, audio, gaming, robotics and business data.
Best for: Rights holders with large libraries of media or business content who want a single marketplace rather than separate deals with each buyer.
Strengths: Troveo reports more than 7,000 licensors who have signed agreements covering AI training, with 95 percent signing exclusively and over $20 million paid out to rights holders so far, according to Troveo. It also describes the wider market: reported deals range from a few million dollars a year for mid-size publishers to $250 million over five years for News Corp and OpenAI, per Troveo.
What to watch for: Troveo's licensing is built around media and content categories. Its figures are self-reported, and the exclusivity share means that a signed licensor gives up rights to others, so read the terms before signing.
3. Opendatabay
Opendatabay publishes a guide to licensed AI training data types, comparing commercial and general licenses for both buyers and sellers.
Best for: Teams that want to understand license types before listing or buying data.
Strengths: Its guide sets out the difference between commercial and general licenses for each side of a deal, which helps owners judge what they are granting, according to Opendatabay.
What to watch for: The guide explains license types rather than showing payouts or timelines, so you will need to ask providers directly about cost, payment and who receives the data.
4. Depositphotos
Depositphotos offers rights-cleared datasets and describes AI data licensing as the legal authorization of datasets used to train machine learning and generative AI models.
Best for: Companies and creators with image or visual libraries that need documented rights before training.
Strengths: Its guide defines AI data licensing in plain terms and explains why rights-cleared data matters for legal risk, according to Depositphotos.
What to watch for: Depositphotos focuses on visual assets. It does not cover internal business records, and it does not publish payout figures in the material we reviewed.
5. Shutterstock
Shutterstock licenses contributor content for AI training through a collective license tied to its contributor fund.
Best for: Photographers, illustrators and video creators who already contribute to a large stock library.
Strengths: A collective license lets contributors take part without negotiating separate deals with each buyer, as Shutterstock explains in its help article on data licensing and the contributor fund.
What to watch for: The option only applies to work already in the library. It is not a route for business records, and the terms of the contributor fund determine what each contributor receives.
6. Mercor
Mercor is one of the expert data providers that Troveo groups with human feedback work for AI labs, according to Troveo.
Best for: Companies that need new expert-generated data or human feedback, rather than licensing existing records.
Strengths: It serves the expert data and human feedback part of the market, which is a different need from licensing archives.
What to watch for: Mercor produces new work rather than licensing records a company already holds, so it does not answer the question of how to earn from existing data. Confirm its terms directly.
7. micro1
micro1 is another expert data provider that Troveo groups with human feedback and expert data companies, according to Troveo.
Best for: Teams that need specialist human input for training or evaluation.
Strengths: It fits the expert and human feedback part of the market, where Troveo places it alongside Surge AI and Mercor.
What to watch for: Like Mercor, micro1 is a source of new expert work, not a way to license archived business records. Check its contract terms before agreeing to anything.
Comparison
| Option | What it licenses | Cost to the data owner | Control over what is shared | Privacy handling | Payout evidence |
|---|---|---|---|---|---|
| Smelted | Internal business records | Free to the company; buyer pays the fee | Company chooses content and buyers | Names, contact details and IDs removed | Ballpark range after matching; figure after sample review |
| Troveo | Media, audio, gaming, robotics, business data | Not covered here | Licensor agreements | Not covered here | Over $20 million paid to 7,000+ rights holders, self-reported |
| Opendatabay | License type guidance for data | Not covered here | Not covered here | Not covered here | Not covered here |
| Depositphotos | Rights-cleared visual datasets | Not covered here | Not covered here | Not covered here | Not covered here |
| Shutterstock | Contributor image and video library | Set by contributor fund terms | Limited to library content | Not covered here | Set by contributor fund terms |
| Mercor | Expert data and human feedback | Not applicable to records | Not applicable to records | Not covered here | Not covered here |
| micro1 | Expert data and human feedback | Not applicable to records | Not applicable to records | Not covered here | Not covered here |
Why clear rights matter
AI training data licensing is the step that makes it clear who may use data and on what terms. Courts have separated how data was used from how it was acquired, and acquisition is where labs have lost ground, including Anthropic's $1.5 billion settlement over pirated books, according to Troveo. Owners who grant a documented license keep a clear record of what they agreed to, which is easier to defend than an informal sale.
FAQ
How do you sell your data to AI companies?
You grant a license to a copy of the data to a buyer, after removing names and identifiers and choosing what is included. Smelted does this for internal business records and introduces the company to two or three matching buyers at no cost to the company. Confirm terms with counsel before signing.
How much money do companies make selling your data?
Amounts vary widely. Troveo reports deals from a few million dollars a year for mid-size publishers to $250 million over five years for News Corp and OpenAI, and Quartz describes that News Corp deal as the largest disclosed AI content licensing agreement. Those are large media deals, not a guide to what a typical company will earn. For Smelted, value depends on headcount, years of records and how connected the data is, and a figure only comes after a buyer reviews a sample.
Is it legal for companies to sell data?
It can be, when the company owns the data or has the right to license it and the contracts are clear. The answer depends on the data, the contracts and your jurisdiction, and no provider can give legal approval on your behalf. Speak with counsel before licensing.
What is AI data licensing?
AI data licensing is the legal authorization of datasets used to train machine learning and generative AI models, as Depositphotos defines it. The owner keeps the underlying data and grants permission for training under agreed terms.
Sources
- Troveo: AI Data Licensing in 2026
- Troveo: Top AI Training Data Companies in 2026, Compared
- Quartz: The price of AI training data, from $5M to $250M
- Depositphotos: Rights-Cleared AI Data Licensing in 2026
- Opendatabay: 2026 AI Training Data Licenses Guide
- Shutterstock: Data licensing and the contributor fund