Data Licensing News
Who is buying AI training data, what they pay, and what the courts just did.Subscribe on Substack

Directory: AI data buyers, marketplaces and vendors

53 public entries across the AI training data market, each checked against a public page on Nov. 16, 2026. Listing is not an endorsement, and we are not paid for placement. The editor has joined or applied to the referral programs of micro1 (Data Payout) and Pewter Data; see the disclosure page.

AI labs that buy data

NameWhat it isSource
AmazonAI content licensee, including a 2025 New York Times deal and later publisher deals. Also a Wikimedia Enterprise API customer, per Ars Technica.Ars Technica
AnthropicFrontier lab. Subject of the Bartz class settlement, finally approved July 20, 2026, at $1.5 billion, per Legal 500. Also named by Wiley as an AI licensing customer.Legal 500
GoogleBuyer of publisher content. In August 2026 it was selected as the $10 million successful bidder for Spirit Airlines' de-identified business data, per the Notice of Auction Results, Dkt. 1463. That is a bid pending court approval, not a completed purchase; the sale hearing was set for Nov. 17, 2026.Notice of Auction Results, Dkt. 1463
MetaBuyer of real-time publisher content for Meta AI and, as reported by The Guardian, a News Corp license of up to $50 million a year.Meta
MicrosoftPublisher marketplace and Copilot licensing, including People Inc. as a launch partner and a 2026 deal with Nine in Australia.TechCrunch
Mistral AIWikimedia Enterprise API customer, per Ars Technica's January 15, 2026 report.Ars Technica
OpenAIFrontier lab and frequent publisher licensee. Public deals in this period include Folha/UOL, Village Media, Times of India parent BCCL, and Indian Express.OpenAI
PerplexityWikimedia Enterprise API customer, per Ars Technica, and a publisher-program licensee in earlier 2025 deals.Ars Technica

Data marketplaces

NameWhat it isSource
BriaGenerative-AI company that signed an opt-in licensing framework with the News/Media Alliance, announced March 23, 2026.News/Media Alliance
Created by HumansBook AI-rights marketplace. Current author-split terms are on the official site.Created by Humans
DataradeGeneral data marketplace.Datarade
Defined.aiData marketplace with a public partnership program. Partner-revenue figures on that page are company claims.Defined.ai
Human Native (Cloudflare)UK AI data marketplace acquired by Cloudflare on January 15, 2026. Terms were not disclosed in the official blog post.Cloudflare
ProRataPublisher licensing intermediary and attributed-answer product (Gist.ai).ProRata
ProtegeReal-world data platform (healthcare and other modalities). a16z announced a $30 million round on January 8, 2026, and said the company already had a $25 million Series A.Andreessen Horowitz
ScalePostPublisher licensing intermediary.ScalePost
SourceXLicenses company data for AI training. Referral terms are on its public how-it-works page.SourceX
TollBitPublisher licensing and crawl-payment intermediary.TollBit
Wikimedia EnterpriseCommercial high-volume Wikipedia API access, not a copyright license (content remains under a Creative Commons license). January 2026 customers included Microsoft, Meta, Amazon, Perplexity, and Mistral, per Ars Technica.Ars Technica

Brokers

NameWhat it isSource
Nyne.aiNyne.ai, which titles its site 'The Data Factory for AI,' with a broker and referral offer. Payout or referral figures on the site are vendor claims.Source
Who Sells DataPublic register of AI training-data and RL-environment vendors, maintained by Secludy AI. It stated '208 vendors ยท updated Nov 11, 2026' on Nov. 16, a vendor-maintained count.Source

RL environment platforms

NameWhat it isSource
Epoch AI (RL environments FAQ)Epoch AI FAQ on reinforcement-learning environments.Source
Isaac LabNVIDIA's open robot-learning framework and simulated environments, used for embodied-agent training.NVIDIA
Ooak DataY Combinator company that describes turning internal company data into a training ground for AI agents.Source
OpenPipe ARTOpenPipe Agent Reinforcement Trainer repository for multi-step agents.Source
Prime Intellect prime-envsEnvironments repository attributed to the Prime Intellect research team.Source
Prime Intellect verifiersPrime Intellect library for RL environments and evals.Source

Referral programs

NameWhat it isSource
Mode.incPublic referral page for company-data licensing. The referral amount is stated on the live page.Mode.inc
Pewter DataPublic site for a referral-led business-data licensing offer. Disclosure: the editor has joined or applied to Pewter Data's referral program; see our disclosure page.Pewter Data

Publishers and rights holders

NameWhat it isSource
ShutterstockContent library with a public history of AI licensing to major labs. Historical 2023-2024 revenue figures predate this window.Shutterstock

Wind-down and shutdown asset sales

NameWhat it isSource
SimpleClosure AssetHubWind-down firm whose AssetHub, per TechSpot on April 19, 2026 (citing Forbes), had facilitated nearly 100 data transactions with reported payouts of $10,000 to $100,000 per company.TechSpot

Trade groups and advocacy

NameWhat it isSource
Authors GuildAuthors' group that publishes explainers on AI training litigation, including the Bartz v. Anthropic settlement.Authors Guild
News/Media AllianceU.S. news-publisher trade body. Announced an opt-in AI licensing agreement with Bria on March 23, 2026.News/Media Alliance
U.S. Copyright OfficeOfficial U.S. Copyright Office hub for copyright and artificial intelligence, including the Part 3 pre-publication report on generative AI training.U.S. Copyright Office

RL training frameworks

NameWhat it isSource
GymnasiumMaintained open-source API for reinforcement-learning environments (Farama Foundation), widely used to train and evaluate agents.Farama
OpenRLHFOpen-source RLHF and RLAIF training framework.OpenRLHF

Open-source lists and repos

NameWhat it isSource
AgentsMeetRL awesome listPublic awesome list for agentic RL.Source
aikorea/awesome-rlClassic public RL awesome list. Older and less frequently updated.Source
awesome-agent-rl-environmentsPublic curated list of training and evaluation environments for LLM and VLM agents.Source
awesome-rl-environments (NafiGit)Public curated list of RL environments.Source
Hugging Face TRLHugging Face library to train transformer language models with reinforcement learning. Released v1.18.0 on Nov. 13, 2026.Source
opendilab/awesome-RLHFPublic curated RLHF resource list.Source
Prime Intellect prime-rlPrime Intellect repository for agentic RL training at scale. Released v0.10.0 on Oct. 13, 2026.Source
verl (verl-project/verl)Open-source RL and agentic training framework (verl-project/verl). Released v0.9.1 on Sept. 20, 2026.Source

Podcasts

NameWhat it isSource
Community Exchange (OpenWeb)Podcast.Source
Latent SpacePodcast on AI engineering and labs.Source
Semafor TechPodcast from Semafor's tech desk.Source

YouTube

NameWhat it isSource
Forbes (YouTube reading of Tong piece)Forbes YouTube video restating Anna Tong's April 16, 2026 reporting on old work Slacks and emails as training data.Source

Want a listing corrected or added? See the corrections policy.