You have until April 24.
After that, everything you type into GitHub Copilot — your prompts, your code, your context, your private repo snippets — gets fed into Microsoft’s AI training pipeline. Automatically. Without asking you first.
From April 24 onward, interaction data including inputs, outputs, code snippets, and associated context from Copilot Free, Pro, and Pro+ users will be used to train and improve GitHub’s AI models unless they opt out. That’s the policy. Buried in a blog post. Sent as an email. Opt out by default.
What They’re Actually Collecting
This isn’t just metadata. The expanded data collection covers inputs, outputs, accepted code snippets, cursor context, comments, file names, repository structure, navigation patterns, and feedback signals like thumbs up and down ratings on suggestions.
Basically everything. Every session. Every prompt. Every piece of code you let Copilot touch.
The interaction data covered by this update may be generated while you are working in a private repository. GitHub says the stored contents of that repository aren’t being used. But the code you’re actively editing with Copilot open? That’s fair game.
Who’s Affected
Copilot Free, Pro, and Pro+ users. That’s most individual developers. Copilot Business and Enterprise users are not affected by this update. Paying organizations get protection. Individual developers get opted in by default.
Spot the pattern.
GitHub’s Justification
They say it makes the product better. GitHub says that as Copilot usage continues to increase dramatically, real world data will help their models cover the increasing number of scenarios they’re now being used for.
They also point out that Anthropic, JetBrains, and corporate parent Microsoft operate similar opt out data use policies. Which is true. And also not really the point. “Everyone else does it” has never been a great argument for doing something without asking.
How to Opt Out Right Now
Go to github.com/settings/copilot and scroll to the Privacy section. Look for “Allow GitHub to use my data for AI model training” and set it to Disabled. Save.
If you have multiple GitHub accounts, make sure to do that for each one.
If you previously opted out of the setting allowing GitHub to collect data for product improvements, your preference has been retained — your data will not be used for training unless you opt in. So if you already turned it off before, you’re good.
Everyone else has until April 24.
The Bigger Picture
This is the trade developers have always been making without fully knowing it. You use a free or cheap tool. The tool gets smarter. You helped train it. GitHub is just being more explicit about it now — which is at least honest, even if the opt-out default isn’t.
The real question is what this data is actually worth to them. GitHub claims models trained on Microsoft employee interaction data showed increased acceptance rates in multiple languages compared to those built solely on public code. Real usage data from 26 million developers is worth a lot more than that.
It’s the same dynamic playing out everywhere right now. Meta is cutting 15,000 jobs to fund AI that will eventually be trained on the work those people used to do. Companies are betting that your output, whether it’s code or content, is more valuable as training data than it is as labor.
If you want AI tools that actually work for you instead of harvesting your work, there are open source alternatives like OpenClaw that run locally and don’t send your data anywhere. And for the AI assistants worth paying for, we put together an honest look at what Claude Pro actually delivers versus the hype.
Your code is valuable. Apparently it’s been helping train AI this whole time.
Now you know. Go turn it off.
