TITLE: GitHub Copilot Will Train on Your Code from April 24 DATE: 2026-03-27 COMPANY: GitHub / Microsoft TOPIC: AI Security SUMMARY: GitHub has announced that from April 24, 2026, interaction data from Copilot Free, Pro, and Pro+ users will be used to train AI models by default. The data collected includes code snippets, accepted outputs, repository structure, and chat interactions. Users must actively opt out via Privacy settings before the deadline. WHAT CHANGED: GitHub announced on March 26, 2026 that it will begin using interaction data from Copilot users to train AI models, effective April 24, 2026. The change applies to users on Copilot Free, Pro, and Pro+ plans. Users on these plans who take no action before April 24 will have their data included in training by default. The data GitHub will collect includes code snippets that are shown to users, suggestions that are accepted, repository structure information, and chat interactions within the Copilot interface. GitHub's parent company, Microsoft, and its affiliates may also receive this data under the updated terms. Copilot Business and Copilot Enterprise users are not affected by the change. These higher-tier plans have historically operated under stricter data protections and the new policy does not alter their terms. The distinction matters for operators: the tiers most commonly used by individual developers and small teams are the ones subject to the change. The opt-out process is available through GitHub account Settings under the Privacy section. Users can disable the option labelled "Allow GitHub to use my data for AI model training." The setting must be updated by each affected user individually. WHY IT MATTERS: The default position is opt-in, meaning any user who does not actively change their settings before April 24 will be contributing data to AI training Businesses that allow developers to use personal or team Copilot Free, Pro, or Pro+ accounts may be unknowingly consenting to client or proprietary code being used as training data Microsoft affiliates receiving the data broadens the potential exposure beyond GitHub's own systems The 28-day notice window is short for organisations that need to go through IT, legal, or compliance review before acting This follows a pattern of AI vendors expanding data use rights as model training costs increase and competitive pressure mounts The policy creates a two-tier system where adequate data protection requires paying for Business or Enterprise plans DAVID & GOLIATH ANALYSIS: GitHub's policy update is a clear signal of the direction the AI tooling industry is heading. The business model logic is straightforward: free and mid-tier users generate interaction data, and that data has real value for improving AI models. The tradeoff is that businesses using these tiers are, intentionally or not, subsidising model improvements with their own code. For lean organisations, the risk is not abstract. A 15-person software consultancy whose developers use personal Copilot Pro accounts may have client code flowing into training data. A product company with a proprietary algorithm may not realise its logic is being used to improve a tool available to competitors. The data is anonymised, but anonymisation is not the same as protection, and the value of training data is in patterns and structure, not in identifying individual contributors. The practical response is straightforward: audit plan tiers, update settings, and document the action. If your business has any material proprietary code or client IP, the cost difference between Pro+ and Copilot Business is likely worth paying for the data protections that come with the higher tier. Do not wait for a compliance review to initiate this conversation. RELEVANT SYSTEMS: Secure AI Brain, Employee Amplification Systems SOURCE URL: https://davidandgoliath.ai/daily-ai-briefing/github-copilot-training-data-opt-out-april-2026 FEED URL: https://davidandgoliath.ai/daily-ai-briefing/feed --- Published by David & Goliath | https://davidandgoliath.ai Daily AI Briefing: one AI development per day, decoded for business operators. This is a structured companion file optimised for LLM retrieval and citation.