Protecting Your Personal Data When Using AI Tools

Chatting with an AI assistant easily feels like a private conversation, so many people unintentionally share information that they would never post publicly. However, what you type may be stored, processed by staff or systems, and in some cases used to improve the service, depending on each provider's policy. This guide helps you build habits for protecting personal data when using AI tools.

Understand what happens to what you enter

Each service handles content differently. Points worth looking up in the privacy policy and terms of use include: how long conversations are kept, whether they are used to train models and whether you can switch that off, who can view the content, where data is stored and how you can delete it. Organisational editions often carry different commitments from free personal versions, so identify which one you are using.

Information you should not enter

A simple rule is to enter only what you would be comfortable having someone else read. Avoid sharing:

ID card or passport numbers, bank account numbers, passwords and verification codes.
Health, financial or private-life information about yourself and others.
Data about learners, customers or colleagues that you are not permitted to share.
Confidential documents of your organisation, unpublished or under confidentiality terms.
Source code containing access keys or internal configuration.

Anonymise before sending

In many cases you can still ask AI for help without identifying information. Before pasting a text, replace names, addresses, phone numbers, emails and ID numbers with generic placeholders such as "[Name A]" or "[Address]". For data tables, remove identifying columns or replace them with random codes. Note that combining several indirect details, such as job title, workplace and a specific event, can sometimes still be enough to identify a person.

Set privacy options and secure your account

Review the service's settings and switch off the option to use conversations for training, if available and if you do not want it.
Use strong, unique passwords for each service and enable two-factor authentication.
Delete conversation history containing sensitive content when it is no longer needed.
Be careful with browser extensions or third-party apps that ask for permission to read what you view or type.
Grant access to files, email or calendars only when truly required for the task.

Be careful with outputs and links produced by AI

Protecting data is not just about what you enter but also what you receive. Do not enter login details on an unfamiliar site just because an AI recommended it. Be careful with links, attachments or code you have not checked. Bad actors can also use AI-generated content to make scams look more convincing, so verify through official channels when a request involves money or personal information.

Follow organisational rules and the law

If you handle other people's data at work or in your studies, organisational rules and laws on personal data protection may apply, even if you only use a tool for temporary processing. Learn your organisation's internal policy on which AI tools are allowed, and ask the responsible unit when unsure.

Conclusion

Using AI safely starts with thinking before you type: read the data policy, do not share sensitive information, anonymise when possible and configure your account carefully. These small habits cost very little effort yet protect you and the people involved from the risk of information leaks.