Does ChatGPT train on your data?
ChatGPT trains on your content by default on Free, Plus and Pro, with an opt-out that binds new conversations only. Business and API do not train by default.
A self-serve opt-out exists
Graded on the plan most people are on. The other plans are in the matrix below.
"As noted above, we may use Content you provide us to improve our Services, for example to train the models that power ChatGPT. Read our instructions (opens in a new window) on how you can opt out of our use of your Content to train our models."
- 21 Sept 2026
- 5 Oct 2026
- Free, Plus & Pro (ChatGPT consumer)
- us
- Fetched from the company
- 8B0D176F…DD49
How we read it: The quote runs to the end of the paragraph. It used to stop after the first sentence, which established the training use but not the opt-out, so this tier's "toggle" rating rested on nothing the reader could check. The second sentence is in the same paragraph of the same capture. The policy doesn't name the setting. It points to an instructions article, and the steps below quote OpenAI's help centre from captures this monitor keeps.
opt out, opt-out, Data Controls, privacy portal, by default, training, Temporary Chat
What the document does offer
- help article, captured 2026-08-24: "You can opt out of training through our privacy portal by clicking on “do not train on my content.”"
- same article, on scope: "Once you opt out, new conversations will not be used to train our models" — forward-looking only, with nothing said about conversations already used
- same article, surviving the opt-out: "If you choose to provide feedback, the entire conversation associated with that feedback may be used to train our models"
- same article, a second control the first does not reach: "adjusting your settings in the ChatGPT interface or privacy portal will not affect these full-environment Codex settings"
- same article: Temporary Chat conversations "won't appear in history, use or create memories, or be used to train our models"
- data controls article, captured 2026-10-01: "Parent or guardian settings and workspace controls may limit whether you can change it."
Why none of it applies: The toggle is real, free and on every consumer plan, which is why this is D and not E or F. The limits sit in the help articles and none in the privacy policy: opting out binds new conversations only, feedback you volunteer is trained on regardless, Codex environments have their own setting that the ChatGPT toggle and the privacy portal both leave untouched, and parent or guardian settings can stop you changing it. A reader of the policy alone would know none of that.
"By default, we do not train on any inputs or outputs from our products for business users, including ChatGPT Business, ChatGPT Enterprise, and the API."
- 10 Sept 2026
- Business, Enterprise & API
- global
- Saved by hand — not re-read automatically
- A8B81A2D…EA89
How we read it: Resolved from UNCLEAR on 24 Aug 2026. The privacy policy defers business data to "customer agreements" this monitor cannot archive, which is why this tier sat unclear. The help article the policy links to answers it plainly and in the opposite direction from the consumer tier: business products do not train by default, and an organization has to opt in. B rather than A because it is an undertaking on a help page, revocable by editing it.
What OpenAI said when we showed them this
Clarified“For services for individuals, such as ChatGPT and Codex, OpenAI may use content to train our models. Users can opt out of model training by either: Turning off “Improve the model for everyone” under Settings → Data Controls in ChatGPT; or Selecting “Do not train on my content” through our privacy portal. [...] After opting out, new conversations will not be used to train our models. [...] For our business offerings, including ChatGPT Business, ChatGPT Enterprise, and the API Platform, OpenAI does not train on inputs or outputs by default.”
What each plan can refuse
The same questions, asked of every plan we could identify. Every cell comes from a quoted clause or a documented search - see the evidence above.
- Self-serve toggle
- Not stated
- 51 / 100
- Not stated
- Unknown
- Nothing to switch off
- Not stated
- 18 / 100
- Not stated
- Unknown
How to opt out of ChatGPT AI training
Plan by plan, from what ChatGPT's own policy says about refusing. Where it publishes the steps, they're listed here with the limits they carry. Where there's nothing to switch off, or nothing graded, this says so instead of inventing a menu path.
Free, Plus & Pro (ChatGPT consumer)
Yes. ChatGPT has a self-serve setting.
- Open your account menu, then Settings.
- Select Data controls.
- Select "Improve the model for everyone", turn it off, and select Done.
Signed in, OpenAI says the setting applies to your account across devices. Parent or guardian settings can stop you changing it. It covers new conversations only: feedback you give (thumbs up or down) can still be used, and Codex's Include environments setting is separate and this switch doesn't reach it.
- Open the sidebar.
- Tap your profile icon to open Settings.
- Turn off "Improve the model for everyone".
- Go to OpenAI's privacy portal at privacy.openai.com.
- Choose "Do not train on my content".
Business, Enterprise & API
Nothing to switch off: training here is off until an opt-in is given.
What this policy has done since we started watching
23 changes detected since 27 Feb 2023. Each one is a diff between two archived copies, and each links to the passage that moved.
- Touched opt-outs, retention, what's collected, plan tiers and training.
- Touched opt-outs, retention, what's collected and training.
- Touched opt-outs, retention, what's collected, sub-processors and training.
- Touched retention.
- Touched opt-outs. Not yet reviewed by a human.
- Touched opt-outs and retention.
- Touched jurisdiction, opt-outs, retention, what's collected, sub-processors and training.
- MajorTouched jurisdiction, opt-outs, retention, what's collected, sub-processors, plan tiers and training.