Guides · Published
Best AI training data licensing companies (2026)
The best AI training data licensing options in 2026, compared on data type, pay, control and privacy. Includes Smelted, Troveo, Opendatabay and Shutterstock.
The strongest AI training data licensing options in 2026 are Troveo for large media and data marketplaces, and Smelted for internal business records from US companies with 50 to 500 staff. Opendatabay and Shutterstock are useful if you want licensed datasets or media rather than your own records.
We make Smelted, so it's on this list. We judged it on the same criteria as the others, and we only describe the other companies using what their own pages say.
How we picked
Each entry is judged on five criteria:
- Data type. What kind of data the option licenses: media, internal business records, or datasets.
- Who gets paid. Whether the rights holder is paid, and how the payment is set.
- Control. Whether the rights holder keeps ownership, chooses what is included, and chooses who receives it.
- Privacy handling. What is removed before data is shared.
- Timeline. How long it takes from first contact to payment, where the company says so.
The list
1. Troveo
Troveo is a licensed data marketplace that signs agreements with rights holders so AI companies can license their content for training. According to Troveo, more than 7,000 licensors have signed agreements that explicitly cover AI training, and over $20 million has been paid out to rights holders so far.
Best for: Rights holders and buyers who need a large catalogue of licensed video, audio, gaming, robotics and business data.
Strengths: Scale is the main point. Troveo says 95 percent of its licensors have signed exclusively, which makes the catalogue easier for buyers to rely on. The company also describes the marketplace as the way to avoid negotiating thousands of separate contracts.
What to watch for: Troveo's scale is built on media and large catalogues. A company with a few hundred thousand internal support tickets or CRM records will fit a different service better.
2. Smelted
Smelted is a free matching service that lists a company's anonymized internal business records for AI data buyers. It finds which buyers fit the company, gives a ballpark range, and introduces the company to two or three matching buyers.
Best for: Founders and C-level leaders at US companies with 50 to 500 employees, founded in 2022 or earlier, who hold years of records in tools such as Zendesk, Jira or GitHub, Salesforce or HubSpot, Bullhorn, AutoCAD or Revit, or Benchling.
Strengths: The company pays nothing, and the buyer pays Smelted's fee. The company keeps ownership, in the form of 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. Core industries are staffing and recruiting, engineering and architecture, B2B software, and science labs and biotech R&D.
What to watch for: Value depends on headcount, years of records and how connected the data is, so there is no fixed price. Payment comes a few weeks to two months after signing, so plan on one to three months from first call to payment. Companies under 30 employees will be routed to a person, because the buyers who take smaller companies pay less.
3. Opendatabay
Opendatabay is a data marketplace that publishes a guide to licensed AI training data types. Its guide compares commercial and general licenses for both buyers and sellers.
Best for: Buyers who want to understand license types before they buy a dataset, and sellers who want to see how licenses are categorized.
Strengths: The guide is a clear reference for the difference between commercial and general licenses. It is written for both sides of a deal.
What to watch for: The guide describes license types. Check each dataset's terms directly before you sign anything.
4. Shutterstock
Shutterstock runs a data licensing program for AI training, with a contributor fund for people whose work is included. Its help article describes the program as a collective AI training license.
Best for: Photographers, illustrators and video creators who want to license existing media collections to AI companies.
Strengths: Contributors are paid through a shared fund, which lets individual creators take part without negotiating their own deal.
What to watch for: This is for visual media, not internal business records. Check the contributor terms before you submit work.
5. Ficstar
Ficstar publishes a list of AI training data providers for 2026. Its list names HumanSignal, which it describes as open source plus managed, covering image, video, text, audio and time series data.
Best for: Teams that need labelled training data and annotation services rather than licensed records.
Strengths: The list gives a quick view of providers across data types, and HumanSignal is named as an open source option with a managed service.
What to watch for: This is a provider list, not a licensing marketplace. Labelling services sell work to AI teams; they do not pay rights holders for their records.
6. Depositphotos
Depositphotos publishes guidance on rights-cleared AI data licensing, covering how deals work and how to reduce legal risk in AI projects.
Best for: Buyers who want to understand rights clearance before licensing images or media for training.
Strengths: The guidance focuses on rights clearance, which matters most when a buyer needs to show where data came from.
What to watch for: The page is guidance. It does not replace legal review, and it does not promise compliance approval.
Comparison
| Option | Data type | Who gets paid | Control for the rights holder | Privacy handling | Timeline |
|---|---|---|---|---|---|
| Troveo | Video, audio, gaming, robotics, business data | Rights holders, over $20 million paid so far | Exclusive licensing for 95% of licensors | Not stated on the page reviewed | Not published |
| Smelted | Anonymized internal business records | Company, via buyer-paid fee model | Keeps ownership as a license to a copy; chooses content and buyers | Names, contact details and IDs removed | One to three months from first call to payment |
| Opendatabay | Licensed datasets | Not stated on the page reviewed | Terms set per dataset | Not stated on the page reviewed | Not published |
| Shutterstock | Visual media | Contributors, through a contributor fund | Contributors opt into the program | Not stated on the page reviewed | Not published |
| Ficstar | Labelled text, image, video, audio, time series | Providers sell annotation work | Not applicable | Not stated on the page reviewed | Not published |
| Depositphotos | Images and media | Not stated on the page reviewed | Rights clearance guidance | Not applicable | Not published |
Where licensed AI data is heading
Demand for licensed data is growing. According to Troveo's statistics page, the global AI training dataset market is projected to reach $16.3 billion by 2033, growing at 22.6 percent a year from 2026, citing Grand View Research. Public text is also running low: Troveo cites Epoch AI's projection that high-quality public text could be effectively exhausted for training between 2026 and 2032, which is one reason labs look for non-public data.
Prices vary widely. Quartz reports that OpenAI's five-year licensing deal with News Corp is worth more than $250 million, the largest disclosed AI content deal. That is a media deal at the top of the market, not a guide to what a mid-sized company can expect.
FAQ
Where can I get AI training data?
You can license it through a marketplace such as Troveo, buy licensed datasets from a provider such as Opendatabay, or license your own internal business records through a matching service such as Smelted. Which route fits depends on whether you are a buyer or a company with records to sell. Troveo describes the marketplace model as the way to reach many rights holders at once.
Are AI companies running out of training data?
Some analysts expect high-quality public text to run short. Troveo cites Epoch AI's projection that it could be effectively exhausted between 2026 and 2032, which is one reason labs are looking for non-public data such as business records.
How do you sell your data to AI companies?
You can list it through a licensing marketplace, or through a matching service. With Smelted, you choose which records are included and which buyers you meet, and names, contact details and IDs are removed before anything is shared. The company keeps ownership and grants a license to a copy. Smelted is free for the company; the buyer pays the fee. Start with the two-minute form at Smelted.
How much is one person's data worth?
It depends on who holds it and how it is used. Smelted licenses company records rather than personal data, so it is set by headcount, years of records and how connected the data is. There is no fixed figure until a buyer has reviewed a sample.
Sources
- Troveo: AI Data Licensing in 2026
- Troveo: AI Training Data Statistics 2026
- Quartz: The price of AI training data, from $5M to $250M
- Opendatabay: 2026 AI Training Data Licenses Guide
- Shutterstock: Data licensing and the contributor fund
- Ficstar: Best AI Training Data Providers in 2026
- Depositphotos: Rights-Cleared AI Data Licensing in 2026
- Smelted
- Smelted: get started