GitHub Will Train AI on Your Code by Default Starting April 24 — Here's How to Opt Out
GitHub announced on March 25 that interaction data from Copilot Free, Pro, and Pro+ users will be used to train AI models starting April 24, 2026, unless users actively opt out. The move affects millions of developers and has triggered a wave of backlash in developer communities.
Starting April 24, 2026, GitHub will use your Copilot interaction data to train its AI models — unless you turn it off first.
The scope of data collection is broader than most users expect. GitHub defines “interaction data” as inputs, outputs, code snippets, and associated context — which in practice means prompts you type, suggestions Copilot generates, code surrounding the cursor, comments and documentation in your files, file names and repository structure, navigation patterns, and any thumbs-up or thumbs-down feedback you’ve given on suggestions. All of it, by default, feeding training pipelines.
Who is and isn’t affected. Copilot Business and Enterprise customers are protected by existing contract terms — their data is not used for training without explicit authorization. Students and educators are also exempt. The change applies specifically to Free, Pro, and Pro+ individual subscribers.
GitHub’s reasoning. The company says it has already trained on interaction data from Microsoft employees and observed “meaningful improvements” — higher suggestion acceptance rates across multiple programming languages. It wants to scale that to paying users. That’s a legitimate argument for model quality. Whether it justifies an opt-out default rather than opt-in consent is the part the developer community is pushing back on.
The backlash is real. GitHub’s community discussion post announcing the change currently has 117 thumbs-down votes. Comments range from privacy concerns to legal questions about code generated in private repositories. Security researchers and open-source advocates have pointed out that European developers under GDPR may have stronger grounds to contest the opt-out framing.
What about private repos? GitHub clarifies that private repository source code stored at rest on GitHub is not used for training. However, interaction data generated during a Copilot session in a private repository — the prompt, the suggestion, the surrounding context — is covered by the new policy. That’s the distinction that matters for teams working on proprietary codebases.
How to opt out. Go to your profile picture → Copilot settings → find “Allow GitHub to use my data for AI model training” → set to Disabled. If you have multiple GitHub accounts, you need to do this for each one individually. If you previously opted out of product improvement data collection, that preference was retained — your data won’t be used unless you actively opt back in.
This fits the broader 2026 pattern. Every major AI company is facing the same pressure: models plateau without new data, and the most accessible data source is their own user base. GitHub at least follows US opt-out norms. Whether that’s sufficient for 40 million developers is a different question.
Sources: GitHub Blog, The Register