Top 20 players in the version control & code hosting space โ Updated June 2026
| # | Platform | Owner | Rating | Founded | Price Tier | Price Details | Comments | Visit |
|---|
| Feature | GitHub | GitLab | Bitbucket | Azure DevOps | Gitea | Forgejo |
|---|
When GitHub Copilot launched in 2021, it was trained on every piece of code that had ever been publicly stored on GitHub โ code written by millions of developers over many years. Much of this code was licensed as open source, meaning people made it public so others could use or learn from it, under specific legal terms. None of those terms authorized a corporation to scrape all of it and build a commercial AI product to sell back to the same developers.
A class action lawsuit was filed in 2022 arguing that Microsoft and GitHub violated open source licenses and the rights of the developers who wrote the code. As of 2026 the legal proceedings are ongoing. The core question the court is deciding: does training an AI on copyrighted or licensed material constitute a violation of that license, or does it fall under fair use? The answer will have enormous implications for the entire AI industry.
Platforms like Gitea, Forgejo, Codeberg, and SourceHut have explicitly stated they do not and will not train AI on user code. GitLab and Bitbucket make it opt-in only. GitHub makes it on by default for all public repositories.