SoylentNews

hubie [soylentnews.org] writes:

GitHub will now use developer data to train its AI models by default [techputs.com]:

GitHub has confirmed [github.blog] it will begin using developer interaction data to train its artificial intelligence models, marking a significant shift in how user data is handled across its platform.

The move, set to take effect on April 24, introduces an opt-out system, meaning most users will be automatically enrolled unless they explicitly disable the setting.
What's Changing

The Microsoft-owned platform said it will start collecting and using interaction data from its AI coding assistant, GitHub Copilot, to improve model performance.

This includes:

• Code snippets entered by users
• Prompts and inputs
• AI-generated outputs and edits
• Context such as file structure and repository data
• User feedback like ratings and interactions

GitHub says this data will help build "more intelligent, context-aware" coding tools and improve accuracy across different programming languages and workflows.

[...] Users who do not want their data used for training must manually disable the setting in their account preferences.

However, enterprise-focused tiers including Copilot Business and Enterprise are excluded from the change, reflecting stricter data governance expectations in corporate environments.

GitHub says real-world developer interactions are essential to improving AI systems.

Original Submission