Does JetBrains AI train on your code? It depends on your license

The Morse teamPublished November 10, 202610 min read

As of November 2026, JetBrains AI does train on your code if your IDE shares what JetBrains calls "detailed code-related data," but only one group has that sharing on by default: individuals on JetBrains' free non-commercial license. JetBrains says this data includes "the full text of inputs sent by the IDE to the large language model and its responses, including source code snippets," and that it can be used for "training AI models" (JetBrains). On paid individual licenses, trials, Early Access builds, and student or open-source licenses, sharing is off until you opt in, and on company licenses it stays off unless an admin allows it.

We read JetBrains' Product Data Collection and Usage Notice (version 1.6, July 7, 2026), its data collection FAQ, the JetBrains AI Terms of Service, its AI service provider list, the non-commercial license agreement, and the AI Assistant, Junie, and Air docs on November 10, 2026. JetBrains deserves credit for how much it publishes. It states a one-year retention period, names every model provider, keeps its Community editions out of the program entirely, and lets you read a log of what AI Assistant sends.

JetBrains AI training: the key facts

Read on November 10, 2026, from JetBrains' own pages.

Does JetBrains train on your code by default?

Only on the free non-commercial license. JetBrains' notice sets the default for each license. For non-commercial, "We may collect analytics as specified above by default based on our legitimate interest. You may opt out in the product settings." For individual commercial, trial, and Early Access licenses, "We start collecting analytics only with your explicit consent" (JetBrains).

The non-commercial license is the free one JetBrains gives individuals for learning, hobby projects, open-source work, and content creation in IDEs including RustRover, WebStorm, Rider, CLion, RubyMine, and DataGrip (JetBrains). Its agreement, version 3.0, effective December 16, 2025, says the product "collects in the default mode certain usage data and data related to your interaction with the Product, including detailed code-related data," and treats using it with collection on as your consent (JetBrains).

JetBrains' FAQ answers "Does JetBrains collect my data to train AI models?" with "Only if you explicitly allow it." A few lines down, it says sharing is enabled if you allow it in the settings "or accept the non-commercial agreement" (JetBrains). On a non-commercial license, accepting the license is the opt-in.

That license also sends anonymous usage statistics with no off switch. JetBrains' FAQ on the free license says of them, "It is not possible to opt out" (JetBrains).

Which JetBrains licenses share code data by default?

We built this table from JetBrains' notice, FAQ, and admin docs, read on November 10, 2026.

Sources: notice, FAQ, and Air admin docs.

Most free licenses start with sharing off. A student or open-source maintainer on a free community license has to opt in, and the free Community editions can't share at all. JetBrains' notice says the unified IntelliJ IDEA and PyCharm downloads, from version 2025.3.1, turn detailed sharing off "when switching to Community" (JetBrains).

What does detailed code-related data include?

More than your AI chats. JetBrains' notice says it covers "terminal usage (command runs) or edit history," plus "complete data about interactions with large language models," and warns that it "may contain personal, business, or project-specific information" (JetBrains). JetBrains' announcement of September 30, 2025 lists "code snippets, prompt text, and AI responses" (JetBrains blog).

JetBrains says the data is used "only for product analytics, model evaluation, and training our own models," and that "We will never sell your data or share it with third parties" (JetBrains). The notice says the data is pseudonymized, stored in the European Economic Area, and not collected in China. The FAQ says JetBrains uses "filtering, anonymization, and strict access controls" to keep sensitive information out of training. It doesn't say what the filters catch, so treat anything you type in the IDE's terminal, including tokens and passwords, as part of the data.

How do you turn off JetBrains data sharing?

In each IDE's settings. Open Settings, then Appearance & Behavior, System Settings, and Data Sharing, and turn off detailed data collection (JetBrains). JetBrains' pages name the checkbox two ways: "Allow detailed data collection by JetBrains AI" in its AI FAQ (JetBrains), and "Send detailed code-related data" in the AI Assistant docs (JetBrains).

Then do it again on every machine and in every IDE. JetBrains' FAQ says "we don't link detailed data sharing settings to your JetBrains Account, so you need to adjust preferences for each device and each JetBrains product you use," and that non-commercial users "will need to opt out again on any new machines they add" (JetBrains). The notice, updated more recently, says preferences "once configured, apply to all installed JetBrains products" (JetBrains). Until the two pages agree, check the setting in each IDE you've installed, on each computer.

What happens to your code after you opt out?

It stays on JetBrains' servers for up to a year. JetBrains' September 2025 post says you can "withdraw consent with immediate effect" (JetBrains blog), which stops new collection. For what you already shared, the FAQ says (JetBrains):

• "Once you stop sharing your data, it stays on our server for one year and is then removed."
• "If you stop sharing your data, previous data could still be used to train the models until the data expires."
• "If the data was already used for training, we may still use the model even after the data is deleted."

The same FAQ says JetBrains "will provide a mechanism to remove data upon request," in the future tense. We found no such mechanism on JetBrains' pages on November 10, 2026. Turn sharing off before you open a project you care about.

Do OpenAI, Anthropic, or Google train on code sent through JetBrains AI?

Not under JetBrains' terms. The JetBrains AI Terms of Service, version 2.1, effective September 30, 2025, say JetBrains will "ensure that all AI Subcontractors commit not to use Your Inputs, Data, Outputs, and Suggestions for the training of their AI models" (JetBrains). JetBrains' AI FAQ adds that it "does not work with the large language model providers that use customer data for training models" (JetBrains).

Those providers still read your code to answer you. JetBrains' provider list, updated September 29, 2026, names OpenAI, Google, AWS, Anthropic, and SpaceXAI for JetBrains AI, and all of them except AWS for Junie (JetBrains). The same list says JetBrains runs Mellum, its own model, and several Qwen models on its own servers, where their "inputs and outputs are not shared with the parties that trained them." The terms also let JetBrains or its subcontractors monitor your inputs for misuse, and say they "may be stored and used for a limited period" (JetBrains).

Older guides cite a JetBrains help page on zero data retention. On November 10, 2026, that page's address redirected to the data collection notice, which doesn't use the term. For what the phrase usually promises, see what is zero data retention.

With your own API key, Junie's docs say it "sends requests to LLMs using your API key directly" (JetBrains), so that provider's terms decide what happens to your code.

Can a company stop JetBrains from collecting its code?

Yes, and that's the default. JetBrains' admin docs say "By default, AI tools and features will not share any code-related data, and your users cannot opt in" (JetBrains). An admin with the Manage AI settings permission can open AI governance, then Settings, and choose Let users decide under Detailed code-related data collection, after which each user picks for themselves.

A company can also sign a contract with JetBrains to share for everyone, and then the setting "applies to all end users and cannot be adjusted individually" (JetBrains). JetBrains pays for that route. Its data collection program offers "complimentary one-year subscriptions for the JetBrains All Products Pack to selected companies willing to share their data," through a waitlist (JetBrains). If your employer joined, the FAQ says you can't change the setting yourself.

How JetBrains' data sharing changed

For how GitHub answers the same question, see does GitHub Copilot train on your code, and how other AI apps compare in which AI companies train on your data.

Where to ask about code without a data sharing checkbox

JetBrains' answer depends on your license, sits behind a checkbox you have to find in every IDE on every computer, and lets JetBrains keep what it already collected for a year after you untick it. On Morse the answer is the same on every plan, there's no checkbox to hunt for, and nothing sits out a year on Morse's servers, because they don't keep your messages, and Morse doesn't train AI on your content. Morse is a web app, so it's for the code you'd otherwise paste into AI Assistant's chat window. Coding, file, and other chat tools are part of the Free plan.

• Private, the mode a new chat opens in, for code from a project you'd rather not see logged anywhere. Morse takes off your account, email, and IP address before DeepSeek, Kimi, GLM, another open-weight model, or xAI's Grok sees the snippet. Morse has zero data retention agreements with its inference, search, and other providers, so the host that answers keeps nothing afterward. When a provider confirms zero retention on each reply, Morse verifies that confirmation and drops any reply that arrives without it.
• E2EE, for code you wouldn't hand to JetBrains, a model host, or Morse. Your browser first verifies the enclave's attestation, then seals the chat for that one enclave, so what crosses Morse's servers is ciphertext Morse can't open. That protection rests on the enclave hardware and attestation working as designed. E2EE runs open-weight models only, without web search or image generation. E2EE is part of the Pro plan, $19 a month, and our E2EE explainer walks through each check.
• Anonymous, when the question needs GPT, Claude, or Gemini. Morse's proxy takes off your account, email, and IP address, and the request reaches the model provider under Morse's account. The provider gets your code without knowing it's you, and it may keep it for a limited time to check for abuse, under its own policy. Put placeholders in for API tokens and internal hostnames first.

In all three modes, Morse doesn't train AI on what you paste, sells no data, and shows no ads, on Free, Pro, and Max alike. Your history sits encrypted in your own browser rather than on Morse's servers, so there's no copy on Morse's servers to wait out. A file you attach travels with that one message, and Morse doesn't store it. Because the history lives on your device, clearing the browser or closing the account deletes those chats, and Morse has no way to restore them, so save anything you'll want later. Morse's privacy policy covers each of these points.

Which JetBrains setting to change, by worry

• Your code training JetBrains' models: on a non-commercial license, turn off detailed data sharing in each IDE, on each machine, before you open the project. On Morse there's nothing to turn off.
• Code you already shared: JetBrains keeps it for up to a year and can still train on it, so the only lever is turning sharing off now.
• Your company's code: leave the organization setting at its default, and check that no one signed up for the data collection program.
• A question about code you can't let anyone read: ask it in an E2EE chat on Morse.

Try a chat from the Morse home page without an account, or start free with no card.