Skip to content

Beginner · Level 1: fundamentals

Your data and privacy

What you never paste into a chatbot, how to find out what a service does with your conversations, and why you stay responsible for what you publish.

9 minUpdated

What you will learn

  • Recognize which data does not go into a chatbot
  • Know where to look for the history and conversation-use settings
  • Anonymize a text before sending it
  • Understand that you are responsible for what you publish

A chatbot feels like a private chat, but it is not a notebook locked in a drawer. What you write reaches a company’s servers, and what happens next depends on the service, the type of account and your settings. So we have a simple rule: if you would not put the information in an email to a stranger, do not put it in a chatbot either.

What you never paste

  • Passwords, authentication codes, access keys and answers to security questions.
  • Banking data: card number, CVV, your IBAN together with your name, account statements.
  • Identity documents: personal numeric code, ID series and number, photos of documents.
  • Other people’s private data: names with address and phone, diagnoses, personal messages, a classmate’s grades.
  • Work secrets: contracts, customer lists, code or internal plans, unless the company explicitly allowed an AI tool.
Careful

Even if a service promises not to use conversations, a password sent once cannot be “taken back”. If you made a mistake and pasted a secret, change that password or key immediately.

Anonymize before you send

You often do not need the real data to get help. Replace names with “Person A”, the company with “Company X”, the exact amount with “a large sum”, the address with “a city in the country”. The result stays useful, and the risk drops a lot.

Tip

A prompt with replaced details: “I have a difficult conversation with a colleague, whom I call Person A. Person A is always late with their part of the project. Give me three polite but firm ways to bring up the subject, each in at most two sentences.” No real name was needed.

What happens to conversations: read the service’s own page

Each service has its own rules about conversation history, about using it to improve models and about how long it is kept, and these differ between free, paid and business accounts and change over time. Do not rely on what you heard; open the service’s privacy or help page and look for the settings.

As an example, on 21 October 2026 Anthropic’s page about its consumer products said that Anthropic may use conversations to improve its models if you choose that or if they are flagged for a safety review, and that “incognito” chats are not used for that purpose. Google’s page for the Gemini apps said that a subset of chats is reviewed by humans and that the “Keep Activity” setting controls whether activity is saved; even with it off, Google still uses chats to respond to you and for protection. Take these lines as an example of how rules differ, not as the rule for these services from today on: pages can change, and in the app, the settings on your own account matter most.

Good to know

The same goes for ChatGPT and any other service: do not assume, read its help page about history, data and deletion. If you work at a company, ask which AI tools are allowed before using them with work information.

Responsibility is yours

If you post, send or submit something written with AI help, you are responsible for the content: whether it is false, whether it violates someone’s rights or whether it reveals personal data. You cannot say “the chatbot wrote it”. If you have questions about personal data, the authority in Romania is ANSPDCP, whose website has information and the complaints procedure. The information in this lesson is not legal advice.

Key takeaways

  • Do not paste passwords, banking data, identity documents, other people’s data or work secrets.
  • Anonymize: replaced names, rounded amounts, general details.
  • Rules about history and training differ and change: read the service’s page and check your own account settings.
  • You are responsible for what you publish or send, even if the AI wrote it.

Try it now

Replace what is in braces with your own details, then copy.

Check what you learned

Three questions. They count for nothing but you.

1/3Which of these pieces of information can you safely paste into a chatbot?
2/3How do you find out what a service does with your conversations?
3/3You published on a blog a text written with AI that contains a false claim about a person. Who is responsible?
Answer all three to check.

Done with the lesson?

Sources