GitHub Flipped Copilot Training to ON the Same Day GPT-5.5 Went GA

Check your GitHub Copilot settings right now.

As of April 24, GitHub's updated privacy policy flipped the default for all Copilot Free, Pro, and Pro+ users: your interaction data, including prompts, suggestions, and code snippets from your context, now trains AI models unless you explicitly opt out. Private repo content at rest is excluded. But interaction data generated while working in a private repo is included. The distinction between "your code" and "your interaction with your code" is doing a lot of work in that policy.

GPT-5.5 went generally available for Copilot the exact same day. Developers who don't opt out are generating training data on OpenAI's strongest agentic model, starting now. At NVIDIA, 10,000 engineers already use GPT-5.5 through Copilot, reportedly cutting debugging cycles from days to hours.

Enterprise and Education tiers are exempt. Microsoft knows corporate counsel won't sign off on training-by-default. Individual developers don't have corporate counsel reviewing their tool settings.

The platform dynamics are straightforward. GitHub has 100M+ developers. The free tier gives OpenAI a massive data flywheel. Developers use the tool. Their usage improves the model. The improved model draws more developers. Same loop Google built with search. The product is the collection mechanism.

"Data may be shared with Microsoft but not third-party model providers." Given Microsoft's OpenAI investment and the coordinated same-day rollout, I'm not sure that line means what it sounds like it means.

Go to Settings, Copilot, toggle off "Allow GitHub to use my data for product improvements." Today.