GitHub will now use Copilot user data for AI training by default, but you can opt out
GitHub has announced that, starting April 24, interaction data from Copilot Free, Pro, and Pro+ users will be used to train and improve Copilot's AI models unless users actively opt out. This change does not affect Copilot Business or Copilot Enterprise users, who remain excluded from the updated data collection program.
For users who previously opted out of GitHub data collection for product improvements, their preferences remain preserved; their data will continue to be excluded from model training. According to GitHub, affected users' interaction data includes code snippets entered or shown to Copilot, outputs accepted or modified, file names, repository structure, surrounding code context, comments and documentation, navigation patterns, interactions with Copilot features such as chat or inline suggestions, and feedback provided on suggestions.
While this scope is broad, GitHub confirms that data from user issues, discussions, or content from private repositories at rest is explicitly excluded. Building on its data practices, GitHub notes that collected information may be shared within its corporate family, including Microsoft, but will not be provided to third-party AI providers. GitHub states these changes align with industry practices and aim to improve Copilot’s code suggestions, security, and bug detection.
- AI Coding Assistant
- Freemium
- Proprietary
Related news
External links
- Updates to GitHub Copilot interaction data usage policyGitHub • Official source
- GitHub Enables Copilot Data Collection For AI Training By Default With Opt-Out SettinggHacks
- GitHub’s Copilot will use you as AI training data, but you can opt outHow-To Geek
- GitHub Copilot will soon use your data to train its model and share it with othersNeowin
- GitHub Copilot will use your data for AI training by default, but you can opt outTechSpot
- GitHub jumps on the bandwagon and will use your data to train AIHelp Net Security
Comments
good strategy to use inactive users' snippets, because they can't opt-out. very ethical. (sarcasm)
And somehow copilot will continue to be the worst LLM on the market. Sad to see popular platforms going against their founding principles and selling out their communities. If only there was a way to reverse ensh*tification
"at rest" means that if any contributor activate Copilot (personal) on a private repository, Copilot will include this repository for its training data.
"Together, we can continue to build AI that accelerates your workflows and empowers you to build better, more secure software faster than ever." It's nice to see Github still having a great sense of humor.