One of the most important privacy controls is the “Improve the model for everyone” setting.

On personal ChatGPT accounts, you can turn this option off if you don't want new conversations to be used to improve OpenAI's models.

Profile → Settings → Data Controls → Improve the model for everyone → Off

When this setting is disabled, your new conversations can remain in your chat history while not being used to train the models.

When you don't want a conversation to become part of your normal chat history or memory, consider using Temporary Chat.

Temporary Chats don't appear in your normal history, don't create memories, and aren't used to improve OpenAI's models while they remain temporary. OpenAI says Temporary Chats may still be retained for up to 30 days for safety purposes.

New Chat → Temporary

This can be useful when you want a more private, non-persistent conversation without changing your account-wide settings.

ChatGPT can use memory and personalization features to make future conversations more useful. However, you may prefer to control what information is remembered.

Review your Memory settings and check what information has been saved. Depending on the features available on your account, you may be able to manage or delete individual memories.

Profile → Settings → Personalization → Memory

Remember that turning off personalization features does not disable safety-related systems that may use limited context in certain situations.

Keeping your ChatGPT history organized is another useful privacy habit. If an old conversation contains information you no longer need, consider deleting it rather than leaving it in your account indefinitely.

This is particularly useful for chats containing personal plans, documents, private discussions, or other information that you don't need to keep.

If you connect external services to ChatGPT, take time to review which accounts and apps are connected. Connected services can provide ChatGPT with access to information needed to complete certain requests, subject to the permissions you have granted.

Only connect services you actually need, and disconnect integrations that you no longer use. OpenAI notes that connected app data can also interact with personalization features depending on your settings.

Sharing a ChatGPT conversation can make it easy for someone else to view specific content. However, you should review old shared links and revoke any that you no longer want available.

If your account provides the shared-link management option, use it to review and remove links you no longer need.

Settings → Data Controls → Shared Links

OpenAI's current documentation describes controls for reviewing and revoking shared links.

Feedback buttons such as thumbs up or thumbs down can be useful for helping improve ChatGPT. However, privacy-conscious users should understand that submitting feedback can affect how the associated conversation is handled.

OpenAI states that even when a user has opted out of model training, choosing to provide feedback can result in the entire conversation associated with that feedback being used to train models.

No privacy setting can replace good security habits. Even if you configure ChatGPT for greater privacy, you should avoid putting extremely sensitive information into a chat unless there is a legitimate reason and you understand how the information will be handled.

For example, avoid sharing:

• Passwords and login credentials
• Authentication or verification codes
• Banking and payment credentials
• Private security questions and answers
• Government identification numbers
• Other information that could create a security risk if exposed

When possible, remove unnecessary personal details before asking ChatGPT to analyze text, documents, or other information.