AI lab

Anthropic: what they buy and what they pay

Anthropic is profiled here as an AI lab[1]. Data types cited in this profile include: Non-public data from third parties, Data from data-labeling services and paid contractors[2].

Last verified November 3, 2026

Quick facts

Quick facts about Anthropic
TypeAI lab[1]
StatusActive[2]
Websitehttps://www.anthropic.com/[3]
HQ548 Market St, PMB 90375, San Francisco, CA 94104, United States (registered address)[3]
FoundedNot published
OwnershipPublic Benefit Corporation; board elected by stockholders and the Long-Term Benefit Trust[1]
ProgramNot published
PriceNot published
Pricing basisNot published
Payment timingNot published
How to startNot published

What they buy

What they buy: Anthropic
Data typesNon-public data from third parties, Data from data-labeling services and paid contractors[2]RL environments (reported)[4]Datasets obtained through commercial agreements with third-party businesses[3]
ExcludedNot published
Intended useModel training[2]

Track record

Named deals

  • Mechanize (RL environments), reported; neither company commented[4]

Coverage

  • TechCrunch (Sept 2025) on labs' push into RL environments, citing The Information on Anthropic's spending discussions (reported)[4]
  • Anthropic's S-1 risk factor (as quoted, Oct 2026): if it cannot rely on fair use, it may be required to pay significant fees to license training data[5]

Watch for

  • $1.5B Bartz v. Anthropic class settlement over acquiring and copying pirated books; final approval July 20, 2026[6]

Sources

  1. Company \ Anthropic, anthropic.com. Accessed October 30, 2026. (company's own site)
  2. Anthropic Transparency Hub, anthropic.com. Accessed October 30, 2026. (company's own site)
  3. Anthropic Privacy Policy, anthropic.com. Published September 10, 2026. Accessed October 30 and November 3, 2026. (company's own site)
  4. Silicon Valley bets big on 'environments' to train AI agents | TechCrunch, techcrunch.com. Published September 16, 2025. Accessed October 30, 2026. (press)
  5. Anthropic’s IPO S1 defends fair use in acquiring & using training data to develop AI (ChatGPTIsEatingtheWorld), chatgptiseatingtheworld.substack.com. Published October 27, 2026. Accessed November 3, 2026. (press)
  6. Court Grants Final Approval of $1.5 Billion Anthropic Copyright Settlement - The Authors Guild, authorsguild.org. Published July 21, 2026. Accessed November 3, 2026. (press)

Last verified November 3, 2026