Guide

AI and your privacy: what you should (and should not) share

Chatbots feel private, but your messages usually travel to a company's servers. Here is how to stay in control.

What can happen to your conversations

• They may be stored for a period of time for safety and operations.
• Some services use them to improve their models unless you opt out - check the data controls in your settings.
• Staff or automated systems may review some chats to catch abuse.
• A hacked account or shared device can expose your history.

A simple rule: if you would not post it publicly, think twice

Avoid sharing: passwords and recovery codes, bank or card numbers, government ID numbers, private medical records, other people's personal details, confidential work documents, and anything covered by a contract or NDA.

Safer habits

• Use the privacy settings: turn off training on your data and use temporary or incognito chats where offered.
• Remove names and numbers before pasting text ("Customer A", "[ACCOUNT]").
• Use work-approved tools for work information.
• Protect your account with a strong, unique password and two-factor authentication.
• Review browser extensions and apps that say they add AI: they may read every page you visit.

Running AI on your own computer

Open-weight models can run locally, so your text never leaves your machine. This takes a reasonably powerful computer but is the most private option.

Children and AI

Talk with children about not sharing real names, addresses, schools or photos with chatbots, and keep an eye on what AI tools they use.

Your rights

Depending on where you live (for example under GDPR in Europe or privacy laws in Canada and parts of the US) you may be able to access or delete data a company holds about you. Look for a privacy request form in the service's settings.