AI coding assistant privacy: is your code used for training?

Updated Oct 26, 2026 · Sources listed at the end

It depends on the vendor, your plan and your settings. Business plans generally exclude your code from training by default; consumer plans may use it unless you opt out. Check the data-usage page of every tool you use. Three examples from vendor documentation:

Claude Code (Anthropic)

• Consumer plans (Free, Pro, Max): you choose whether your data may be used to improve future models; when the setting is on, Anthropic trains on data from those accounts, including Claude Code use.¹
• Commercial terms (Team, Enterprise, API, third-party platforms): Anthropic does not train generative models on code or prompts sent to Claude Code unless the customer opts in.¹

Cursor

Privacy Mode can be turned on in settings or by a team admin, on free or Pro plans; when it is on, Cursor says it will not train on your data and has contractual controls with model providers.²

GitHub Copilot

GitHub’s documentation describes the training data behind Copilot’s models as publicly available text and code, and warns that suggestions can occasionally match code in that training set.³ GitHub also offers a Data Protection Agreement supporting GDPR compliance.⁴

Questions to ask any vendor

1. Is my code used for training by default on my plan?
2. How long are prompts and code retained?
3. Which model providers see my code, and under what terms?
4. Can an admin enforce the setting for the whole team?

Tools mentioned

• Claude Code: paid from $20/mo
• Cursor: free plan, paid from $8/mo
• GitHub Copilot: free plan, paid from $10/mo

Sources

1. Claude Code docs: Data usage
2. Cursor: Security (Privacy Mode)
3. GitHub Docs: Responsible use of Copilot inline suggestions
4. GitHub Copilot plans