GitHub Copilot Will Train AI Models on Your Code Starting April 24
GitHub announced a major policy change that will affect millions of individual developers using Copilot. Starting April 24, 2026, interaction data from Copilot Free, Pro, and Pro+ users will be used to train and improve AI models unless users explicitly opt out.
What Is Changing
From April 24 onward, GitHub will collect interaction data from individual Copilot users to train its AI models. This includes inputs, outputs, code snippets, and associated context from your coding sessions. The data collection applies to Copilot Free, Pro, and Pro+ subscribers.
Notably, Copilot Business and Copilot Enterprise users are exempt from this change. Enterprise customers will not have their interaction data used for model training, maintaining the privacy expectations that come with organizational licenses.
What Data GitHub Will Collect
The interaction data that may be used for training includes:
GitHub emphasizes that this data will not include content from your issues, discussions, or private repositories at rest. However, they note that Copilot does process code from private repositories when you are actively using the service, and this interaction data could be used for training unless you opt out.
How to Opt Out
If you prefer not to participate in this data collection program, you can opt out through your GitHub settings. Navigate to your Copilot settings and look for the Privacy section. Users who previously opted out of data collection for product improvements will have their preferences preserved automatically.
Opting out does not affect your access to Copilot features. GitHub confirms that users who choose not to participate will still be able to take full advantage of the AI features they know and love.
Why GitHub Is Making This Change
GitHub explains that real-world interaction data leads to smarter models. Their initial models were built using publicly available data and hand-crafted code samples. Over the past year, incorporating interaction data from Microsoft employees has yielded meaningful improvements, including increased acceptance rates across multiple programming languages.
The company believes that training on diverse real-world interaction data will help models better understand development workflows, deliver more accurate and secure code pattern suggestions, and improve their ability to help developers catch potential bugs before they reach production.
Data Sharing and Privacy
The data used in this program may be shared with GitHub affiliates, including companies in their corporate family such as Microsoft. However, GitHub states that this data will not be shared with third-party AI model providers or other independent service providers.
Industry Context
This policy update aligns with established industry practices for AI model improvement. Many AI coding assistants use similar approaches to enhance their models through real-world usage data. The change represents GitHub's effort to close the gap between their training methodologies and those of competitors who have already been leveraging user interaction data.
FAQ
Do I need to do anything if I use Copilot Business or Enterprise?
No. This policy change only affects Copilot Free, Pro, and Pro+ users. Business and Enterprise customers are not affected by this update.
Will opting out reduce Copilot's functionality?
No. GitHub explicitly states that users who opt out will still have full access to all Copilot features. The opt-out only prevents your interaction data from being used to train AI models.
What happens if I do nothing?
If you take no action and are a Copilot Free, Pro, or Pro+ user, your interaction data will automatically be used for model training starting April 24, 2026. If you previously opted out of data collection, your preference has been retained.
Can I opt out after April 24?
Yes. You can change your privacy settings at any time through your GitHub Copilot settings. Opting out after April 24 will stop future interaction data from being used for training.
Does this affect code in private repositories?
Code from private repositories at rest is not included. However, when you actively use Copilot, it processes code from private repositories to provide suggestions, and this interaction data may be used for training unless you opt out.
