GitHub to Use Copilot Data for AI Training from April 24

GitHub, the Microsoft-owned code hosting platform, announced on Wednesday that it will begin using interaction data from its AI-powered coding assistant, Copilot, to train and improve its models starting April 24. The change applies to users on Free, Pro, and Pro+ plans, who will be automatically included unless they opt out through their account settings.

Under the updated policy, GitHub will collect data generated during Copilot interactions, including user prompts, generated code suggestions, accepted or modified outputs, code context, comments, documentation, file names, repository structures, and feedback on suggestions. This information helps the platform analyze development workflows, enhance code accuracy, and detect potential issues more effectively. The company emphasized that data from private repositories, issues, or discussions stored at rest will not be used for training. Only data from active sessions with Copilot, such as inline suggestions or chat features, will be eligible if the user has not opted out.

Copilot Business and Enterprise users, as well as those with enterprise-owned repositories, are exempt from the change. Individuals who previously disabled data...