Copilot individual plans default to training use
Since April 24, 2026, GitHub may use Copilot interactions from Free, Pro, Pro+, and Max individual plans to train AI models unless the user opts out. Eligible data can include prompts, outputs, code context, comments, filenames, and repository structure.
Published 24/4/2026
What happened?
GitHub introduced an opt-out training-data policy for individual Copilot plans, effective April 24, 2026. Eligible interaction data includes inputs, outputs, code snippets, surrounding context, comments, filenames, repository structure, navigation, feature use, and feedback. GitHub says private-repository content at rest is excluded, but interaction data processed during active Copilot use can be used unless the user opts out. Business, Enterprise, enterprise-owned repository data, and opted-out users are excluded.
Why it matters
A freelancer can use a personal Copilot subscription inside a private client repository and still transmit interaction data eligible for model training. Repository visibility is therefore not the relevant control; account plan, settings, client terms, and the material sent to the assistant are.
What should you do?
Before using a personal plan on client work, review the engagement’s confidentiality and AI terms. If training use is not expressly acceptable, disable “Allow GitHub to use my data for AI model training” in Copilot settings. Avoid sending secrets or unnecessary proprietary context regardless of the setting.