Services
Explore
Free AI assessmentAnswer 15 questions about one process and see where AI could help.
Start an assessmentDoes ChatGPT train on my data?
It depends on the service. This guide separates consumer ChatGPT, business plans and the API, and explains how training differs from retention and access.
Intro
Intro
On consumer ChatGPT (free and Plus), your conversations may be used to improve OpenAI's models unless you turn that off in Data Controls. ChatGPT's business plans are covered by separate business terms, so check the current terms for your plan. For the API, OpenAI's documentation states that API inputs and outputs are not used for training by default unless the customer opts in. Retention, access and what the service stores are separate questions, and each varies by service.
The answer depends on which service you use. Free and Plus are consumer plans. Team and Enterprise are business plans with separate terms. The API is a developer service with its own terms, and it is not a ChatGPT subscription tier.
Training is one part of the picture. Turning training off does not decide how long OpenAI keeps your data, who can access it, or what the service stores, such as chat history and memory. This page treats each of those separately.
Per-plan breakdown
How ChatGPT's plans and the API treat your data for training
The main distinction is between consumer plans (free and Plus), business plans (Team and Enterprise) and the API. Plus is a consumer plan, so paying for it does not change the training terms.
Plan
Price
Used for training?
Terms
Human review
Notes
| Plan | Price | Trains On Data | Retains30Days | DPA | Human Review | Notes |
|---|---|---|---|---|---|---|
| Free | £0 | opt-out | possible | Used for training unless you turn it off in Data Controls. Check the current retention for deleted chats. | ||
| Plus | Check OpenAI's current pricing | opt-out | possible | Consumer terms, as for the free plan. Plus adds usage and features. It does not move you onto business terms. | ||
| Team | Check OpenAI's current pricing | no | restricted | Business terms. Check the current terms on training, retention and data processing agreements. | ||
| Enterprise | Check OpenAI's current pricing | no | none | Business terms with admin controls. Retention and access depend on the agreement and the settings you choose. | ||
| API | Charged by usage | no | restricted | OpenAI's API documentation says API data is not used for training by default unless the customer opts in. Retention varies by endpoint and feature. Needs technical setup. |
Consumer plans: used for training unless you turn it off
Free tier: turn off in Data Controls
ChatGPT Plus: turn off in Data Controls
Business plans and the API: not used for training by default
ChatGPT Team
ChatGPT Enterprise
OpenAI API
How to turn training off
Turning off training on the free and Plus plans
These steps apply to the consumer free and Plus plans. Team, Enterprise and API customers should check their own terms and settings. Menu names can change, so check OpenAI's help centre if the wording differs.
Select your profile name or icon in ChatGPT, then choose Settings.
In Settings, open Data Controls, where the training setting sits.
Turn off the setting that lets OpenAI use your conversations to improve its models. OpenAI has labelled it Improve the model for everyone. The change applies to future conversations.
Deleting chats and turning off memory are separate settings from training. Each affects what ChatGPT stores, so check what each one does in OpenAI's help documentation.
Close and reopen Settings, and confirm the setting is still off.
What "training" means
Training, retention and access are separate questions
Training on your data describes one use among several. Knowing which use is involved helps you judge the exposure for your own work.
PlainPeople rate the model's responses, and the model is adjusted to produce more of what rates well
What It MeansConversations used for model improvement may be shown to a reviewer who rates the response. This is one reason the human review column matters.
PlainExamples of good conversations are used to teach the model format and style
What It MeansConversations, or edited versions of them, can serve as training examples. If one of yours is selected, what you wrote could influence how the model responds to similar inputs.
PlainLarge amounts of text from the internet and other sources are used to build the model's general knowledge
What It MeansTreat any conversation used for model improvement as something that could influence future models. OpenAI's documentation describes how it uses conversations for each service.
PlainConversations can be checked, automatically and sometimes by people, for misuse
What It MeansThis is separate from training. Turning off training does not turn off safety monitoring. Data kept for monitoring is held for periods set out in OpenAI's documentation, and those periods differ by service.
Common questions
FAQs
Does ChatGPT train on my data by default?
On the consumer free and Plus plans, conversations may be used to improve OpenAI's models unless you turn this off in Data Controls. Business plans have separate terms, so check the current terms for your plan. For the API, OpenAI's documentation says API data is not used for training by default unless the customer opts in.
If I turn off chat history, does that stop training?
Chat history, memory and training are separate settings, and OpenAI has changed how they relate over time. Check OpenAI's current help documentation for what each setting does. Whichever settings you choose, OpenAI keeps some data for a period for purposes such as safety monitoring.
Can I delete my existing conversations to stop them being used for training?
Deleting a conversation removes it from your history. OpenAI keeps deleted conversations for a period before removing them from its systems, as set out in its current documentation. You cannot find out from OpenAI whether a particular past conversation was used in training. Turning off training applies to future conversations and does not remove past data from models already trained.
Does ChatGPT Plus stop training on my data?
No. Plus is a consumer plan. It adds usage and features, and the training setting works as it does on the free plan: conversations may be used unless you turn it off. Business plans have different terms. Check the current terms for each.
Could my conversations be used to train future models?
On consumer plans with training turned on, OpenAI's terms allow conversations to be used to improve its models, and that can include future versions. If you turn training off, or use a business plan or the API under terms that exclude training by default, check the current documentation to confirm how that applies to your account.
Is the API different from chatgpt.com for training?
Yes. The API is a developer service with its own terms, separate from ChatGPT subscriptions. OpenAI's API documentation says API data is not used for training by default unless the customer opts in. Retention and what is stored vary by endpoint and feature, so check the documentation for the features you use. The API also needs technical integration.
What does 'training' actually involve: does a human read my conversations?
Training mostly processes large volumes of data automatically. OpenAI also uses people to review a sample of conversations for quality and safety and to label training data. Check OpenAI's privacy documentation for who can access conversations on your plan and under what conditions.
Talk to us
We can help you choose a ChatGPT setup and settings for each type of work
We compare plans and settings against how your team uses AI, and can draft an internal AI use policy for your team and advisers to review. Questions about your obligations under UK GDPR need a qualified adviser.