What are the privacy concerns with AI chatbots?
The main AI privacy concern is that, depending on the service and your settings, chats may be stored, reviewed by people or used to train future models. So never share passwords, codes, bank or ID numbers, your address, private health details or other people's personal information, and keep work data out unless your employer approves the tool.
▶ Try it
Why it matters
Chatbots feel private, but what you type may be stored, read by reviewers or used for training. You'll learn what to keep out, what to anonymise and which settings to check.
In 6 minutes you'll be able to
- Passwords, codes, ID numbers and bank details never go into a chat.
- Remove names and details before asking about health, money or other people.
- Check the training, history and sharing settings of every chatbot you use.
What you'll practise
Share, anonymise or keep out?
You're about to type these into Lumi. Where does each one belong?
- "Give me a 3-day study plan for a maths test" → Fine to share
- Your email password, so Lumi can "help sort your inbox" → Keep out
- Your passport number for a visa form → Keep out
- A message from your friend Jade, with her name and phone number → Anonymise first
- Your blood test results with your full name on them → Anonymise first
- Your company's unreleased sales figures → Keep out
- "Explain what a credit score is" → Fine to share
Lumi's privacy settings
Many chatbots have settings like these. Set each one the privacy-friendly way.
- Use my chats to improve the model
- Delete chat history after 30 days
- Remember personal details across chats
- Anyone with the link can see this chat
- Sign-in with two-step verification
Safe chatbot habits
Choose every habit that protects your privacy.
- Replace names with placeholders like "Person A"
- Read the privacy settings when you start a new service
- Paste a whole document with everyone's contact details
- Ask general questions instead of sharing your own records
- Assume deleted chats are always gone at once
- Use only work-approved AI tools for work data
Remember
If you wouldn't post it publicly, don't paste it into a chatbot.
If you wouldn't post it, don't paste it.
Key facts
- 81% of Americans who have heard of AI say the information companies collect will be used in ways people are not comfortable with, and 70% have little or no trust in companies to use AI responsibly. · Source: Pew Research Center, How Americans View Data Privacy, 2023
- The US Federal Trade Commission has warned AI companies that breaking privacy promises, or quietly using people's data to train models, can violate consumer protection law. · Source: FTC, "AI Companies: Uphold Your Privacy and Confidentiality Commitments", January 2024
- Depending on the service and your settings, chatbot conversations may be stored, reviewed by people for quality or safety, or used to train future models.
- In the UK and EU, data protection law gives you rights over your personal data, including the right to ask a company for a copy of what it holds about you.
Updated:
Questions people ask
Do chatbots store your conversations?
Many do, at least for a while. How long, who can see them and whether they're used for training depends on the service and your settings, so check its privacy page.
Is it safe to share health information with AI?
Be careful. Remove your name and identifying details, ask general questions, and talk to a doctor or pharmacist about anything that matters for your health.
Can I stop a chatbot from training on my chats?
Many services let you switch off the use of your chats for training in the settings. Look for options about improving the model, chat history or data controls.
Can I use a chatbot for work documents?
Only if your employer allows it, and only with approved tools. Confidential, customer or personal data usually needs a tool your company has checked.
Related terms