Data Licensing Directory

You have data. Where can you license it?

Machine Observer maps public routes from a data owner to AI-data marketplaces, brokers and supplier programs. Every listing names the provider source and observation date. Routes are not rankings or endorsements.

provider-sourced evidenceowner-side routingevidence-backed routesmachine-readable API
Start with what you have

Common paths into the market

These groupings are Machine Observer-derived from the data types each provider publicly says it handles. They are filters, not quality scores.

Route finder

Company email, chat, documents and operational history

These providers publicly describe licensing business records, proprietary datasets, operational data or broad multimodal/text data.

Route finder

Audio, video, transcripts and creator-owned archives

These providers publicly describe routes for audio, video, speech, creator media or multimodal datasets.

Route finder

Scientific, engineering, code, spatial or specialized proprietary datasets

These providers publicly describe specialized, technical, code, spatial, physical or proprietary data programs.

Commercial routes

5 current owner-side routes

Observed 2026-10-01. Acceptance, buyer demand and commercial terms can change after the observation date.

Defined.ai

What it publicly accepts: audio · video · images · text · multimodal · coding · STEM

Who it appears built for: Businesses that generate or aggregate data, plus professional data suppliers.

Process: Apply as a data partner; Defined.ai describes quality, licensing-rights, legal, compliance and cybersecurity review before marketplace participation.

Commercial structure: Dataset licensing through its marketplace; commercial terms depend on the dataset, usage and customer needs.

Defined.ai says there is no fixed minimum dataset size; niche datasets may still qualify.Supplier participation is subject to vetting and documented rights/compliance requirements.

Source: Defined.ai Partnership Programs ↗

↗

Fiund

What it publicly accepts: audio · video · transcripts · creator-owned archives · UGC

Who it appears built for: Creators and rights holders with real-world audio, video, transcripts or owned media archives.

Process: Contributor onboarding and valuation, followed by licensing through Fiund.

Commercial structure: Fiund publishes contributor terms that give it exclusive AI-licensing representation while describing buyer licenses as non-exclusive.

Exclusive representation is different from transferring ownership; review the contributor agreement before onboarding.Fiund states that contributors keep ownership and may withdraw content from future licensing.

Source: Fiund — License your library ↗

↗

License Your Data

What it publicly accepts: business email · chat · documents · spreadsheets · PDFs · operational files

Who it appears built for: Operating businesses; the provider states 20+ employees, 3+ years in business and years of real email or chat as its standard qualification checks.

Process: Qualification, storage sizing, buyer bid, de-identification and export if the business accepts.

Commercial structure: Provider says there is no upfront cost and it is paid from a completed licensing deal.

Published deal sizes, buyer participation and minimum-bid claims are provider claims and are not independently verified by Machine Observer.Business records may contain third-party confidential, personal or regulated data; ownership of a file does not automatically establish the right to license every datum inside it.

Source: License Your Data ↗

↗

Protege

What it publicly accepts: healthcare · video · audio and speech · text · spatial data · motion capture · 3D · other proprietary datasets

Who it appears built for: Data and content providers across regulated and non-regulated domains.

Process: Protege describes discovery, licensing structure, data preparation and inclusion in qualified AI-data deals.

Commercial structure: Revenue share / partner compensation through licensed use by AI model builders.

Protege emphasizes rights, provenance, privacy and dataset preparation before delivery.Acceptance and economics depend on buyer demand and the provider’s review.

Source: Protege — For Data Providers ↗

↗

Troveo

What it publicly accepts: business records · documents · work histories · CRM · finance and operations · code · video · audio · gaming · robotics

Who it appears built for: Businesses and rights holders with proprietary operational histories or content catalogs.

Process: Troveo describes inventory, rights review, privacy review, packaging and buyer licensing.

Commercial structure: Provider says owners keep ownership and approve licensing terms; licensing may be non-exclusive or exclusive depending on the agreement.

Troveo’s payout, buyer-count and rights-holder-count figures are provider-published claims and are not independently verified by Machine Observer.Operational datasets often require privacy, contract and third-party-rights review before licensing.

Source: Troveo — Business Data ↗

↗
Broader market map

Reference infrastructure

Useful when the five owner-side routes above do not fit what you have.

DataLicenses.org ↗

Evidence-oriented catalog of licenses, marketplaces, preference signals, registries, access controls, standards and governed sharing approaches for AI data use. It is a market map, not a transactional licensing broker.

60Current initiatives reported by source
18Marketplaces reported by source

Source observed 2026-10-01: DataLicenses.org ↗

Fiund AI Data Licensing Index ↗

Provider-hosted research dataset of publicly reported AI content and data licensing agreements. Fiund states that inclusion does not imply a customer, supplier, buyer or partner relationship.

32Public deals reported by source
62Source links reported by source
16Deals with reported economics reported by source

Source observed 2026-10-01: Fiund AI Data Licensing Index ↗

Use the directory

Find a practical route for your dataset.

Start with the free matcher to narrow the market by data type and owner profile. For a more specific machine-readable answer, preview a route plan or a provider dossier before deciding which providers to investigate.

Evidence boundary

What Machine Observer is asserting

That these public routes exist and that the providers describe the listed data types, eligibility or process on the cited pages as of the observation date.

Provider claims about buyers, payouts, deal sizes or likely value remain provider claims unless Machine Observer obtains independent evidence.

Before sharing data

Rights are dataset-specific.

Owning a mailbox, drive or archive does not automatically establish the right to license every item inside it. Contracts, confidentiality, privacy, personal data, client materials, employee communications, releases and sector rules can all change what is licensable.

This directory is market infrastructure research, not legal advice or a determination that a particular dataset may be licensed.