How do I opt out of Claude training on my data?
Go to claude.ai, open Settings, then Privacy, and turn off the 'model improvement' toggle. Your consumer chats are used for training only if that setting is on, if a chat is flagged for safety, or if you opt in. Use Incognito for chats never used for training. API and Claude for Work data aren't trained on by default.
Why — the first-principles explanation
On Anthropic's consumer plans (Free, Pro, Max), whether your chats help train future models is a setting you control, not an automatic fate. The company will use your chats to improve models only if you allow it, if a conversation is flagged during a safety review, or if you explicitly opt in. So 'opting out' really means making sure the allow-setting is off.
That control lives in Privacy Settings as a Model Improvement toggle. Turn it off and your ordinary conversations won't be pulled into training. For a chat you never want used, even if that toggle is on, use Incognito mode, which is excluded from training by design. There's a retention angle too: if you do allow training, your data can be kept for an extended period (up to 5 years); leaving it off keeps you under Anthropic's standard, much shorter consumer retention.
Businesses have an even cleaner path. Commercial products, the API and Claude for Work, are not used to train Anthropic's models by default, which is why sensitive or regulated data belongs there rather than in a personal consumer account. The takeaway: check the toggle, use Incognito for anything sensitive, and route work data through commercial products.
An example that makes it click
It's like the 'help improve this app by sharing usage data' checkbox you see when installing software. If it's checked, your activity gets studied to make the product better; uncheck it and it doesn't. Claude's Model Improvement toggle is that checkbox for your chats. And Incognito mode is like using a notepad you shred afterward, nothing from it ever goes into the 'improve the product' pile, even if the checkbox is on elsewhere.
How to do it
- Open claude.ai and sign in, then click your name or the menu to reach Settings.
- Go to the Privacy section.
- Find the 'model improvement' or 'help improve Claude' toggle and switch it off.
- For sensitive conversations, start an Incognito chat, which is never used for training.
- For work or regulated data, use the API or Claude for Work, which aren't used to train models by default.
Key facts
- Consumer chats are used for training only if you allow it, if flagged for safety, or if you explicitly opt in.
- The opt-out is the Model Improvement (help improve Claude) toggle in Privacy Settings, set to off.
- Incognito chats are never used to improve Claude, even with Model Improvement on.
- If you allow training, data can be retained up to 5 years; otherwise standard shorter retention applies.
- API and Claude for Work data are not used to train Anthropic's models by default.
Anthropic's assistant, known for long-document reasoning and careful, safe answers.
Affiliate link — we may earn a commission at no cost to you.▶ The 60-second explainer (script)
How do you opt out of Claude training on your data? On the consumer plans, it's a setting you control. Go to claude.ai, open Settings, then Privacy, and find the model improvement toggle, sometimes called 'help improve Claude', and switch it off. That's the opt-out. By design, your chats are only used for training if that setting is on, if a conversation gets flagged for safety, or if you specifically opt in. For anything really sensitive, use Incognito mode, which is never used for training even if that toggle is on. One more thing worth knowing: if you do allow training, your data can be kept up to five years, so leaving it off keeps retention short. And if you're handling work or regulated data, use the API or Claude for Work, those aren't used to train Anthropic's models by default. Check the toggle, use Incognito, done.
What authoritative sources say
People also ask
Where exactly is the opt-out setting?
In claude.ai under Settings, then Privacy: the Model Improvement (help improve Claude) toggle. Turn it off to opt out.
Does opting out delete my past chats?
Opting out stops future training use and shortens retention going forward. Delete specific conversations separately if you want them removed.
Is there a way to guarantee a chat isn't used?
Yes, use Incognito mode. Those conversations are never used to improve Claude, even if Model Improvement is on.
What about my company's data?
Use the API or Claude for Work. Commercial products are not used to train Anthropic's models by default.