AI and Privacy — What You Should Know

Here’s what you should know to protect sensitive information.

What does AI see?

When you type to ChatGPT, Copilot, or another AI, the text is sent to a server. The AI:

  • Reads everything you write — questions, texts, files
  • Can save conversations to improve the service, depending on the service’s settings and terms (at least in most free versions)
  • May use content to train language models (depending on the service)

What shouldn’t you share?

Never share this with AI:

  • Social security numbers, passwords, or bank details
  • Health information (diagnoses, medications, test results)
  • Work emails or confidential work documents
  • Other people’s personal data without permission
  • Trade secrets or contracts

Rule: If you wouldn’t tell it to a stranger on a bus, don’t tell it to AI.

How do you limit data use?

In most AI services, you can turn off conversation history:

ChatGPT: Settings → Data Controls → “Improve the model for everyone” → off

Copilot: Check the privacy and history settings in Copilot and delete conversations you don’t want saved.

Gemini: Settings → Activity → you can turn off history

Note: free versions often collect more data than paid ones.

AI and GDPR

If you use AI at work in the EU, GDPR applies:

  • Don’t enter customer data without a legal basis for processing
  • Your employer may have their own rules for AI use
  • Ask your employer which service is allowed

Practical tips

  1. Anonymize — replace names and identifiable information
  2. Use a work account — your workplace’s paid account is often more private
  3. Clear your history
  4. Read the privacy policy — even if it’s boring, it tells you what happens to your data

Summary

AI can be used safely when you know the boundaries. Most everyday use — like brainstorming, writing, and learning — carries no significant privacy risk, as long as you avoid sharing sensitive information.


Previous: AI for Teachers Next: ChatGPT vs Copilot vs Gemini