Data Privacy & Personal Safety
How to Check Privacy Settings and Chat History in AI Apps
हिंदी में पढ़ेंPrivacy settings help you control parts of an AI app's behaviour, but different controls do different jobs. Review training, history, memory, sharing, deletion and permissions separately. Do this before entering personal information, and recheck after significant app updates or account changes.
For example, a parent switches off model improvement and expects old chats to disappear. They remain visible because training and chat history are separate controls. Reading what each switch changes avoids a false sense of privacy.
Where should I start in any AI app?
Open the official app or website and look in the account settings for Privacy, Data controls, History or similar headings. Confirm that the account is yours and check whether a school or workplace manages it.
Read the current help page for each setting. Look for options to delete conversations, export information, manage shared links and review connected services. A missing switch is not evidence that the app never stores data.
What can I check in ChatGPT?
In the current ChatGPT guidance, Settings → Data controls includes “Improve the model for everyone.” Turning it off keeps new conversations from being used for training under that control; the chats can still appear in history. Menu wording and availability can change. AI-S04: Read the source
Review memory or personalisation separately if available. Do not assume disabling training erases remembered details, previous conversations, shared links or information already sent elsewhere. Use the official instructions for deletion rather than guessing from a button label.
How does Temporary Chat differ?
OpenAI says Temporary Chats are not used to improve models while they remain temporary, but a copy may be kept for safety for up to 30 days. Saving changes the chat's status, and current options may use existing personalisation when chosen. It is not a guarantee that no data is processed or retained. AI-S05: Read the source
Treat temporary mode as one control, not a safe place for passwords, identity documents or private child photos. A setting cannot make every possible upload appropriate.
Which device settings matter too?
Review camera, microphone, location, notifications and photo-library permissions in the device or browser settings. Allow only what the activity needs. Turn off permissions that no longer serve a purpose and keep the app updated.
Try this with a fictional practice chat first: locate it, check whether it was shared and find the deletion option. Do not use sensitive data to test a privacy feature. If a child is involved, an adult should operate the service where required by its age rules and explain the choices in simple words.
Where can you check the details?
What can you read next?
Want to check the sources?
These notes explain what was checked for the article. The date is an information check, not a publication date.
- AI-S04: Data controls in ChatGPT
OpenAI
Where to look: Settings > Data controls; training versus history
How it was checked: Official text retrieved
Controls vary by account and workspace. Training opt-out does not delete history.
- AI-S05: Temporary chat in ChatGPT
OpenAI
Where to look: Temporary chat, personalization, saving and retention
How it was checked: Official text retrieved
While temporary, not used for model improvement; safety copy may be kept up to 30 days. Saving changes its status; existing personalization may be used when chosen.
Next planned check: