Privacy settings to change in your AI assistant first
Most chat assistants use your conversations to train future models unless you tell them not to. The switch is usually a few taps away, but the name and the menu differ in every app. Below are the paths as they stood when we re-checked them on October 14, 2026, along with what each provider keeps after you delete a chat.
What "training on your chats" means
When a provider trains on your conversations, people and systems at that company may review what you typed, and the patterns can shape later versions of the model. Turning training off does not delete your chat history; it stops new chats from being used that way. Deleting a chat is a separate step with its own timelines.
Work accounts often follow different rules. Content from ChatGPT Business, Enterprise and Edu is not used for training by default, and Grok's business and enterprise content isn't either. If your employer gave you an account, ask which plan it is before you paste work files into it.
ChatGPT
Training is on by default for personal Free, Plus and Pro workspaces. Go to Settings, then Data controls, and turn off "Improve the model for everyone". Your chat history stays. Temporary chats are never used for training but may be kept for up to 30 days for safety. Watch the thumbs: rating an answer up or down can let that whole conversation be used even after you opt out. Deleted chats, memories and accounts leave OpenAI's systems within 30 days, with legal and safety exceptions.
Pi
Inflection's privacy policy says conversations may be used for training and that you can opt out on your Account settings page. Memories can be viewed and deleted under Settings, then Account, then Manage Memories. For Pi users, inputs are generally kept for at most 15 days after you delete them.
DeepSeek
DeepSeek's terms let you opt out by turning off "Improve the model for everyone"; you can also email its privacy team. Keep in mind that account data stays for as long as you have an account, and that data is stored in the People's Republic of China. The Government of Canada restricted the app on its own mobile devices in February 2025.
Grok
In the app, open Settings, then Data Controls, and turn off "Improve the model". On grok.com the path is Settings, then Data, then "Improve the Model". Private Chat conversations are not used for training and are deleted within 30 days. Deleted conversations and accounts are also removed within 30 days, unless kept longer for legal or safety reasons.
Poe and 
Jasper
Under Poe's privacy policy, the third-party companies whose models you use may receive your chats, including uploaded files, and may use them to improve their services. We found no Poe setting to stop that. Temporary chats are deleted within 24 hours, and Memory is off unless you turn it on.
Jasper's Terms say it uses your inputs and outputs to improve its services and develop its own models. Only users in the EU and UK get a stated right to object, so Canadian users have no published opt-out.
Habits that matter more than settings
- Never paste your SIN, passwords, bank details or health records into any assistant.
- Replace names with initials before you paste a document.
- Use temporary or private chat modes for anything sensitive.
- Re-check your settings after big app updates; menus move.
- Delete old chats you no longer need.
Canada's privacy regulators have looked closely at this market. In May 2026 they found that ChatGPT's initial training did not comply with their privacy laws, and in June 2026 the federal commissioner found that Grok's image tool launched without proper safeguards. Settings help, but the safest data is the data you never typed.
For the full picture on each provider, see the rankings, or read our guide to checking answers.