GitHub
@github
📣 Important updates on how we'll use interaction data to improve GitHub Copilot. Get details and learn how to manage your preferences here. https://t.co/JOIuZQxzXr
29retweets113likes
View on XGitHub to Train Copilot Models on Individual User Interaction Data by Default
GitHub· Updated
Starting April 24, GitHub will use interaction data from Copilot Free, Pro, and Pro+ users to train its AI models unless they manually opt out. This shift moves beyond public datasets to incorporate real-world developer workflows, including code snippets and repository structures, to improve model accuracy.
GitHub announced that interaction data from Copilot Free, Pro, and Pro+ tiers will be used for model training starting April 24, 2026. This includes
inputs, outputs, and code snippets. While the service already processes this data to function, it will now contribute to model improvement by default for individual subscribers.This update shifts model training toward proprietary interaction data to refine performance. GitHub reported that incorporating data from Microsoft employees has already increased code acceptance rates. By expanding this to the broader user base, the models can better understand diverse development patterns and suggest more secure code.
You can manage these preferences in the GitHub settings under the Privacy section. Users who previously opted out of data collection for product improvements will remain opted out. Notably, this policy change does not apply to Copilot Business or Copilot Enterprise users, whose data remains excluded from training.
Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →
