AI training opt-out
Definition
AI training opt-out is a privacy setting that stops an AI provider from using your conversations to train its future models. In consumer chat apps it is usually a toggle in the account's data or privacy settings.
By Ronan Liedmeier, fractional head of growth and course author · Published 17 October 2026 · Last updated 17 October 2026
Also called: model improvement setting, data training toggle
It matters whenever you paste in client documents, internal numbers or anything covered by an NDA. Defaults differ between products and between personal and company accounts, so check the setting rather than assume it.
How it works
In ChatGPT the setting is called "Improve the model for everyone", under Data controls. OpenAI says that when it is off, new conversations are not used to train its models, and it states that it does not train on content from its business workspaces by default. In Claude's consumer apps the setting is Model Improvement, in the privacy settings. Anthropic says it uses consumer chats for training when you allow it there, and may also use chats flagged for safety review to train safety systems. In both products, temporary or incognito chats are excluded from training.
In ChatGPT, turning the setting off does not delete your existing chats. Feedback you give with a thumbs up or down can be used for training in both products, and OpenAI says this applies even after you opt out.
Example
A consultant wants an assistant to summarize a client contract that falls under an NDA. First she checks that her account's training setting is off, then she replaces the client's name and the fees with placeholders. The summary is just as useful and the identifying details stay on her laptop.
Common mistake
Treating the opt-out as permission to paste anything. It covers training only: the provider still stores your chats and may review some for safety. Keep a short never-paste list (passwords, ID numbers, other people's private data), and use a work account your company has approved for client material instead of a personal one, which is how shadow AI starts.
Learn it in AI Academy
- Privacy, Safety & Data HygieneBeginner · 9 min
Related terms
- AI memory (chat memory)
- AI memory is a chatbot feature that carries details from past conversations into new ones, such as your job, preferences or ongoing projects, so you don't have to repeat them. You can usually view, edit or delete what it has stored, or turn it off.
- Shadow AI
- Shadow AI is the use of AI tools at work without the company's approval or knowledge, such as staff pasting client documents into a personal chatbot account. It creates data risk, and it also shows where people already find AI useful.
- AI governance
- AI governance is the set of rules, owners and checks an organization uses to decide which AI uses are allowed, how their risks are handled, and who answers when something goes wrong. In the US, NIST publishes a voluntary framework for it, the AI Risk Management Framework.