GitHub to Use Personal Copilot Interaction Data for Model Training
GitHub Copilot is an AI coding assistant that generates suggestions based on user prompts, code and project context. Its personal plans include Free, Pro and Pro+. The policy change affects how developers’ source code and interaction data may be reused, highlighting the trade-offs among model performance, data governance and privacy protections.
Starting April 24, 2026, GitHub will by default be permitted to use inputs, outputs, code snippets and related context from Copilot Free, Pro and Pro+ users to train and improve AI models. Copilot Business, Copilot Enterprise and eligible users receiving educational benefits will not be affected. Personal-plan users who object must proactively disable the data-training option in their settings.
All Coverage
1 original reportsThe Backstory
The history behind this eventGitHub Copilot Tightens Individual Plan Limits, Pauses New Sign-Ups
GitHub Copilot has expanded from a code-completion tool into an AI agent capable of running lengthy, multitask workflows, driving up token usage and inference costs. GitHub says some requests cost more to process than an entire month's subscription fee. That suggests fixed-price plans such as the $10-a-month Pro tier and $39-a-month Pro+ tier can no longer support heavy agent workflows while maintaining reliable service.
On April 20, 2026, GitHub paused new applications for Copilot Pro, Pro+ and Student plans, while Free remained unaffected. It also tightened session caps and token allowances calculated every seven days, with the Pro+ cap more than five times that of Pro. Pro users can no longer access Opus models, and existing paying subscribers may request refunds through May 20.
Subscribe to Mark Radar Weekly
Every Friday, the week's strongest signals in your inbox. Unsubscribe anytime.