Understanding AILesson 5/8

Lesson 5 of 8 · 10 min

Privacy and security

What happens to your messages and what you should never share.

Writing to an assistant feels like a private conversation. In reality, every message goes to a company’s servers. That is no reason to avoid it, but it is a reason to know what to share, and what to keep to yourself.

What happens to your messages

When you send a message to an assistant such as ChatGPT, Claude or Gemini:

  • it is sent and stored on the company’s servers, often in your history;
  • it may be reviewed by the company’s teams, for example to check for abuse;
  • depending on your settings and account type, it may be used to improve future models.

Most assistants offer settings to limit this: opting out of your conversations being used for training, deleting your history, or opening a temporary chat that is not kept in your history. Business accounts generally offer stronger guarantees.

What you should never share

Even with good settings, some information has no place in a conversation with an AI:

  • passwords, verification codes and answers to security questions;
  • credit card, bank account or social insurance numbers;
  • photos of ID documents: driver’s license, passport, health card;
  • health information about someone else;
  • data about your customers or colleagues: names, contact details, files;
  • your employer’s confidential documents: contracts, internal figures, unannounced projects.

A simple rule: do not write anything you would not be comfortable seeing on a screen in a public place.

Anonymize before you send

Good news: the assistant almost never needs the real information to help you. Replace it before sending.

  • “Marie Tremblay, 514-555-0199” becomes “the customer”.
  • A company name becomes “Company A”.
  • Exact amounts become rough figures, unless they are the heart of the question.

The result will be just as good, and you add the real information yourself at the end, in your email or document.

AI used against you

Fraudsters use AI too. Three habits to know:

  • Fake emails are better written. A lack of typos no longer proves a message is genuine. Be wary of urgency and links, and go through the official website.
  • Voices can be imitated. A call from a “loved one” in distress asking for money? Hang up and call the person back on their usual number. Agree on a secret family password.
  • Images and videos can be fabricated. A shocking or too-good-to-be-true video deserves a check before you share it.

Choose safe tools

Use the official apps and websites of assistants. Be wary of unknown browser extensions and “free” apps that promise AI: some read everything you type.

Key takeaways

  • Your messages are stored on the company’s servers: check your privacy settings.
  • Never share passwords, card numbers, ID documents or your customers’ data.
  • Anonymize before sending: the assistant does not need real names to help you.
  • Be wary of fraud made more convincing by AI: verify through another channel.

Quiz

1. What happens to a message you send to an online assistant?
2. Which of these should never be shared with an assistant?
3. You want help replying to a customer. What is good practice?
4. A “loved one” calls you in a panic and asks for money right away. What do you do?
Next lesson: Ten useful everyday uses
Next