26 March 2026, The Register in an article titled "GitHub: We going to train on your data after all" reported that GitHub will, starting 24 April 2026, by default use code and content in public repositories to train its AI models — the report summed it up colloquially, "As of April 24 you'll be feeding the Octocat unless you opt out".
The report noted that the policy is not without an opt‑out mechanism: GitHub says repository or account owners can prevent their content from being used for training via an official "opt‑out" control. GitHub describes this as a user‑controlled choice, but the initial post did not provide full technical details or screenshots of batch‑operation tools, and The Register emphasized the tension created by the timeline and the default setting itself.
The change touches several specific points of contention. First, the default‑on setting means many developers who are unaware or who do not adjust settings in time will have their code absorbed into training data; second, questions about open‑source license applicability have resurfaced: some maintainers worry whether training use conflicts with project license terms, especially repositories with "non‑commercial" restrictions or more stringent contributor agreements.
The report also described a fast and varied community reaction: some maintainers have started urging peers on social platforms to check repository settings and consider collective action, while security researchers warned that large‑scale training could increase the chance models learn dangerous or noncompliant code, raising new compliance and security challenges.
From GitHub/Microsoft's perspective, The Register quoted their intention to improve the intelligence and usefulness of developer tools such as Copilot by enlarging training sets to improve suggestion quality and coverage. GitHub stressed that the change does not target private repositories and will not automatically bypass content designated as protected or untrainable — but the article saw the "public repositories included by default" stance as a shift in position.
Legal and governance issues were also a focus: some open‑source legal experts interviewed by The Register said that although many licenses do not explicitly prohibit training use, default training shifts choices that contributors previously controlled about distribution scope, potentially prompting new litigation or regulatory needs. This is particularly sensitive for maintainers who rely on special license terms to sustain their project ecosystems.
The article urged developers to do two things: first, watch for GitHub's forthcoming details on the specific operational paths and API support and set account or repository preferences in time; second, review project licenses and contributor agreements and, where necessary, discuss with the community whether to adjust licenses or publish explicit statements. The Register summarized in a pointed tone that this policy is both a result of technological evolution and a test of platform governance and community trust.
For IT developers and open‑source contributors, the article advised two practical lines: check and decide repository training permissions before 24 April 2026, and follow ongoing discussions about licenses and model use — these choices will directly affect how future models "see" and reuse the code you write.
▶️ Like this article? Creating content is hard!
Follow our YouTube channel:
AI Rosetta
Your likes, subscriptions and shares are our biggest support! 🙏
More AI news at: https://www.airosetta.com