GitHub will use Copilot interaction data to train models by default — opt-out available

GitHub announced on March 25, 2026 that it will begin using interaction data from Copilot Free, Pro, and Pro+ accounts to train and improve its AI models by default, effective April 24, 2026. Individual users who do not want their Copilot sessions used for model training can opt out in their personal Copilot settings.

What data may be used

GitHub says the interaction data that may be collected for model training can include a broad range of information from Copilot sessions. Examples listed by the company include: - Inputs and prompts you send to Copilot - Outputs and suggestions Copilot provides, including suggestions you accept or modify - Code snippets and the code context around your cursor - Comments and documentation you write - File names, repository structure, and navigation patterns - Chat and inline interactions with Copilot - Thumbs up/thumbs down feedback

Who is not affected

Copilot Business and Copilot Enterprise accounts are excluded from this default training policy; GitHub also notes student and teacher exemptions. GitHub says it will not pull code “at rest” from private repositories for training. However, code and surrounding context that are actively sent to Copilot during a session can be collected for training unless the user has opted out.

How to opt out

Individual users can disable model-training collection in their personal Copilot settings. GitHub’s instructions point to: Settings → Copilot → Features → Privacy → “Allow GitHub to use my data for AI model training” and set it to Disabled. The company also says prior opt-out choices will be preserved, so users who already disabled the setting should not need to reconfigure it.

Data access and safeguards

GitHub states that interaction data used for training may be accessed by GitHub and Microsoft personnel and shared within GitHub affiliates, and that it is not sold to third-party model providers. The company describes automated filtering for secrets and additional access controls and audit logging as safeguards around the data. Community and press coverage confirmed both the policy change and the location of the opt-out setting.

What this means for developers

If you use Copilot Free, Pro, or Pro+, your Copilot sessions may be included in GitHub’s training data starting April 24 unless you explicitly opt out. Users who are concerned about their session data being used for model training should check their Copilot privacy setting and disable the training toggle if they do not want their interactions included.

Bottom line: GitHub is moving to make Copilot interaction data available for model training by default for individual Free/Pro/Pro+ users, but provides an opt-out switch in personal settings and excludes business/enterprise plans and code at rest in private repositories.