Do AI coding tools train on your code? Settings by plan
When you use an AI coding tool, your prompts and bits of your code go to the vendor. A common question is whether that vendor can use them to train its models. The answer depends on the tool and on your plan, and the rules change. This guide sums up what each vendor's own page said when we read it on October 15, 2026. Plan details are marked checked Oct 2026.
Not affiliated with or endorsed by GitHub, Anysphere (Cursor), Anthropic, OpenAI, Google, Windsurf or JetBrains. Check each vendor's current terms before relying on these details.
This isn't legal advice. Your own contract, your company's agreement and your admin settings decide what applies to you. Check them, or ask your admin.
Our tool pages: GitHub Copilot · Cursor · Claude Code · OpenAI Codex
Quick table (checked Oct 2026)
| Tool | Free / individual plans | Business / team | Enterprise | Where the setting is |
|---|---|---|---|---|
| GitHub Copilot | Free, Pro, Pro+, Max: may be used for training since April 24, 2026; you can opt out | Not used for training | Not used for training | GitHub → profile picture → Copilot settings → "Allow GitHub to use my data for AI model training" |
| Cursor | Not used unless you agree, flag/feedback exceptions; Privacy Mode stops training. Default: not stated on Cursor's pages. Turn on Privacy Mode in settings to be sure. | Admin can enable Privacy Mode; members inherit it | Same as team | Cursor settings (Privacy Mode) |
| Claude Code (Anthropic) | Free, Pro, Max: only if you allow it; you choose at signup | Team: not trained on under commercial terms | Enterprise and API: not trained on under commercial terms | Claude Privacy Settings |
| OpenAI Codex / ChatGPT | Free, Plus, Pro: may be used; you can opt out | Business: not used by default | Enterprise and Edu: not used by default | ChatGPT Settings → Data controls → "Improve the model for everyone"; Codex has a separate "Include environments" setting |
| Gemini Code Assist | Individual tier stopped serving requests June 18, 2026 | Standard: prompts and responses not used for training | Enterprise: same | No training setting described |
| Windsurf | Policy says it depends on your terms; check your plan's terms. | Check your terms | Check your terms | Check your terms |
| JetBrains AI | Non-commercial license: detailed code data may be collected by default, opt-out | Organization license: off by default, admin must allow, then user consents | Same as organization | IDE product settings |
GitHub Copilot
GitHub's Copilot policy docs say that starting April 24, 2026, on Copilot Free, Pro, Pro+ or Max, "GitHub may use your interactions with GitHub features and services—including inputs, outputs, code snippets, and associated context—to train and improve AI models." The same page describes it as an opt-out, so it's on unless you turn it off.
- To turn it off: click your profile picture on GitHub, then Copilot settings, then set "Allow GitHub to use my data for AI model training" to Disabled.
- Business and Enterprise: "GitHub does not use Copilot Business or Copilot Enterprise customer data to train AI models." If you don't see the setting, GitHub says to check that you aren't signed in with a Business or Enterprise license.
Cursor
Cursor's privacy policy says: "We do not use Inputs or Suggestions to train our models, or permit third parties to use them for training, unless" they're flagged for security review, you report them (for example as feedback), or "you've explicitly agreed to their use for such training purposes."
Its security page adds that Privacy Mode "can be enabled in settings or by a team or enterprise admin," is "available to anyone (free or Pro)," and "When enabled, we will not train on your data." New team members inherit the team's setting.
- Default: Not stated on Cursor's pages. Turn on Privacy Mode in settings to be sure.
Claude Code (Anthropic)
Anthropic's Claude Code data usage docs split plans in two:
- Free, Pro and Max (consumer): "We will train new models using data from Free, Pro, and Max accounts when this setting is on (including when you use Claude Code from these accounts)."
- Team, Enterprise, API and third-party platforms (commercial): "Anthropic does not train generative models using code or prompts sent to Claude Code under commercial terms," unless the customer chose to share data, for example through the Development Partner Program.
- Anthropic's commercial privacy article adds that feedback you send (such as thumbs up or down) may be used for training.
- Choosing and changing it: Anthropic's consumer terms update says these rules apply to "Claude Free, Pro, and Max plans, including when they use Claude Code" from those accounts, and not to services under its Commercial Terms or API use. New users choose "during the signup process," and you can change it "at any time in your Privacy Settings." If you allow training, data retention is five years.
- Incognito: Anthropic's consumer privacy article says: "Your Incognito chats are not used to improve Claude."
OpenAI Codex and ChatGPT
OpenAI's help article on model improvement says:
- Free, Plus and Pro: "we may use your content to train our models." To opt out, "turn off Improve the model for everyone under Settings > Data controls in ChatGPT, or select Do not train on my content in our Privacy Portal." It says either one covers "ChatGPT conversations and Codex tasks."
- Codex environments: "Codex has a separate Include environments setting in Codex settings for allowing training on full environments." Changing your ChatGPT setting doesn't change it.
- Feedback: on individual plans, if you give a thumbs up or down, OpenAI says "we may use the entire conversation associated with that feedback to improve our models, even if you have opted out of model training."
- Business, Enterprise, Edu and API: "By default, we don't use inputs or outputs from ChatGPT Business, ChatGPT Enterprise, ChatGPT Edu, or our API to improve our models." OpenAI's enterprise privacy page says the same: "By default, we do not use your business data for training our models."
Gemini Code Assist (Google)
- Standard and Enterprise: Google's data governance page says: "Gemini doesn't use your prompts or its responses as data to train its models."
- Individual tier: Google's deprecation page says that starting June 18, 2026, the IDE extensions "stopped serving requests for the Gemini Code Assist for individuals, Google AI Pro, and Google AI Ultra tiers." We didn't cover the replacement products.
Windsurf
Windsurf's privacy policy says training on user content depends on "the terms that apply to your use of the Services." We couldn't find a Windsurf page that clearly sets out defaults by plan, so we aren't guessing.
JetBrains AI
JetBrains' data collection and use policy covers "detailed code-related data," which includes prompts and code snippets. It says this data "can be used for product improvement and training AI models."
- Non-commercial license: it may be collected "by default," and you can opt out in the product settings.
- Commercial individual, trial or EAP: collection starts "only with your explicit consent."
- Organization license: it's "disallowed by default and can only be enabled by a company admin."
Quick steps
- Find your plan in the table above.
- If you're on a free or individual plan, check the setting named for your tool.
- If you're on a business plan, ask your admin what's switched on, and check your contract.
- Keep secrets out of your project folder. A line in your AGENTS.md or CLAUDE.md file asking the agent not to open secrets files can help, but it isn't a guarantee.
Not affiliated with or endorsed by GitHub, Anysphere (Cursor), Anthropic, OpenAI, Google, Windsurf or JetBrains. Check each vendor's current terms before relying on these details.