GitHub moves Copilot training on individual plans to an opt-out default
Original: Updates to GitHub Copilot interaction data usage policy View original →
Hacker News pushed GitHub’s March 25, 2026 policy update to 303 points and 143 comments because it touches the default bargain behind AI coding tools: better models in exchange for more user interaction data. The post matters less as a technical launch and more as a governance signal. GitHub is saying that, for individual Copilot tiers, real-world coding interactions are now important enough to become part of the default training path unless users explicitly turn that path off.
The change starts on April 24 and applies to Copilot Free, Pro, and Pro+ users. GitHub says the interaction data in scope includes prompts, outputs, code snippets, and surrounding context. The company also says Copilot Business and Copilot Enterprise are not affected, and that users who had already disabled data collection for product improvement will keep that preference. GitHub frames the decision as a way to make Copilot more context-aware and more useful across a broader set of development workflows.
One detail that likely drove the HN reaction is GitHub’s justification. The company says it has already seen measurable gains after training with Microsoft employee interaction data, including better acceptance rates across multiple languages. That gives the announcement a practical, product-operations tone rather than a research tone: GitHub is effectively arguing that hand-curated datasets and public code are no longer enough if it wants Copilot to improve on real developer habits, edits, and failure cases.
The tradeoff is straightforward. More interaction data may help GitHub improve suggestion quality, bug detection, and workflow awareness, but it also sharpens long-running questions about privacy, consent defaults, and what developers are actually comfortable sending back to a model provider. Hacker News reacted to exactly that tension. The policy does not ban opting out, but it makes the default posture for individual users meaningfully different from business customers, which is why the announcement landed as more than a routine settings change.
Original source: GitHub Blog
Related Articles
GitHub said in a March 17, 2026 X thread that Copilot coding agent now adds model selection, self-review before PRs, built-in code/secret/dependency scanning, custom agents, and cloud-to-CLI handoff. GitHub’s blog frames the upgrade as a smoother delegation workflow for background coding tasks.
GitHub said on March 20, 2026 that Copilot code review has surpassed 60 million reviews. The company’s March 5 blog says usage is up 10x since launch, now covers more than one in five code reviews on GitHub, and relies on an agentic architecture tuned for higher-signal feedback.
GitHub has launched a public preview that lets teams assign Jira issues directly to the Copilot coding agent and receive AI-generated draft pull requests in GitHub. The company says the integration reduces context switching while preserving existing review and approval controls.
Comments (0)
No comments yet. Be the first to comment!