Can you sell your GitHub data for AI training?
GitHub's terms let the account holder export its own data, and we found no clause in its API terms about training AI. The content itself carries low rights risk. This is research from public sources, not legal advice.
- The holder's own export
- The holder can export it
- Data pulled through its API
- No AI clause found
- Rights risk of the content
- Low
What a holder should know
GitHub's terms are not the blocker; the content is. A repository mixes the holder's own code (sellable only if IP assignment from employees and contractors is clean) with vendored third-party and open-source code whose licences travel with it, plus issues, pull request comments and commit metadata containing contributors' names and emails (personal data). Public repositories already carry a licence to other users under D.5. Since 24 to 27 April 2026, GitHub may use Copilot inputs and outputs on Free, Pro and Pro+ to train AI models unless the user opts out (J.3), and this does not apply where a GitHub Customer Agreement governs; disclose this to a buyer seeking exclusivity over code written with Copilot. Commentary (Heise, March 2026) reports private repository content at rest is not used for training.
Getting your own data out
Any owner can clone repositories. Organisation owners can run migration exports. No plan gate on getting your own code out.
GitHub Terms of Service (Effective date: April 27, 2026), D.3: "You own Your Content." Repositories can be cloned in full with git, and organisation and user migration archives are available to owners.
Apps that use its API
No clause found in the GitHub Terms of Service or Acceptable Use Policies barring a repository owner from licensing its own code for AI training. The API terms (section H) restrict abuse: "Abuse or excessively frequent requests to GitHub via the API may result in the temporary or permanent suspension" and "You may not use the API to download data or Content from GitHub for spamming purposes." Acceptable Use, section 7: "Researchers may use public, non-personal information from the Service for research purposes, only if any publications resulting from that research are open access." Scraping "does not refer to the collection of information through our API."
The sources, word for word
- GitHub Terms of Service(effective 2026-04-27)
“You own Your Content.”
- GitHub Terms of Service, J.3 (Inputs and Outputs)(effective 2026-04-27)
“a license to collect and use your Inputs and Outputs to develop, train and improve artificial intelligence ... unless (a) you opt out through your account settings”
- GitHub Acceptable Use Policies(read 2026-10-11)
“Researchers may use public, non-personal information from the Service for research purposes, only if any publications resulting from that research are open access.”
- Heise: Only with opt-out, GitHub will train Copilot models with user data(2026-03-27)
“GitHub is changing its terms of use and will use user interactions with Copilot for its AI model training starting April 24, 2026. ... The change affects the Free, Pro, and Pro+ plans, but not Business and Enterprise.”
Research from public sources, read on the day shown. It is not legal advice: a sale needs a lawyer who has read the records and the contracts.
What GitHub holds that buyers ask for
Source code with its history
Repositories with commits, reviews and the tickets they answer
Project and ticket histories
Jira, Linear, ServiceNow or Asana records that show how work moved from request to done
Who buys it
- Turing (Lazarus)
"A single payout of up to $1M" for an operating record. For code repositories its process page (lazarus.turing.com/process, read 16 Oct 2026) says "Repos typically pay $10k–$100k+ depending on quality and complexity".
- Ooak
"Earn up to $1M" (raised from $300K by 15 Oct 2026). Deals to date run from $20K to several hundred thousand dollars.
- Labelbox (Prism)
"Smaller targeted datasets starting around $50K and large enterprise partnerships reaching $1M+".
- Troveo
"AI labs and startups are actively paying six figures for company data exports", and Troveo pays the holder "on every sale, not just the first".
- Corpus Bridge
It cites licences of about $10,000 to several hundred thousand dollars for ordinary companies.
- SimpleClosure AssetHub
"Tens to hundreds of thousands of dollars for unique assets". Early deals were reported at $10,000 to $100,000.
What your GitHub records are worth, and what to settle before anyone sees them.
In this topic
Rights and law
- Start hereCan this data legally be sold or licensed for AI training?Privacy law, copyright and fair use, software vendors' terms, de-identification, bankruptcy sales and what a licence agreement holds.
- GuideIs It Legal to Sell Customer Data? A US Guide for AI TrainingOften yes, but through layers: your privacy policy, the FTC Act, state and sector laws, and customer contracts. De-identified records are the main lawful path.
- GuideFair Use and AI Training: What US Courts Have RuledUS courts are split: training was fair use in Bartz v. Anthropic and Kadrey v. Meta, not in Thomson Reuters v. Ross, and pirated copies lost. Where each stands.
- GuideIs It Legal to Scrape Data for AI Training? (US Law, 2026)Scraping public pages is rarely a federal crime after Van Buren and hiQ, but contract, copyright, anti-circumvention and privacy law apply. Resale is hard.
- GuideCan a Company License Employee Emails and Slack for AI Training?Usually yes for business records the company controls, but only after it handles monitoring notices, third parties in the messages, privilege and vendor terms.
- GuideSelling Customer Data in Bankruptcy: Section 363 ExplainedA bankrupt company can sell customer data only if the sale fits its privacy policy, or a court-appointed privacy ombudsman reports and the judge approves it.
- GuideWhat Is a Data Licensing Agreement? The Terms for AI TrainingA data licensing agreement lets a buyer use data for set purposes while the holder keeps ownership. Each clause AI buyers ask for, and where holders stand.
- GuideHIPAA De-Identification for AI Training: Safe Harbor vs Expert DeterminationHealth data leaves HIPAA only through Safe Harbor (remove 18 identifiers) or an expert's determination of very small risk. How each works for AI training.
- GuideLicensing Company Data for AI Training Under the GDPR (2026)Yes, an EU, UK or Swiss company can license its records for AI training, if it has a lawful basis, tells people, drops sensitive data and uses a transfer tool.
- GuideSelling Data for AI Training: Privacy Law by CountryOutside the EU and UK, licensing records for AI training mostly needs consent or anonymisation, and a contract or consent to send them to a US buyer.
- RegisterLawsuits and rulingsEvery case, ruling and settlement over training data.
- RegisterLicence termsWhat buyers ask for in a licence, and where holders stand.
- PageDe-identification standardOur rules for removing people from records, each with its source.
- RegisterAI crawler censusA daily census of robots.txt across the most visited sites.