Keeping student details safe when using AI tools

The simplest rule is the best one. If a detail would identify a child or a parent, don't type it into a general AI chat. You can nearly always get the same draft with a label like "the student" instead.

What to keep out

Treat these as off limits for any public chat tool.

• Names of students and parents, together or separately.
• Phone numbers, email addresses and home addresses.
• Dates of birth and school names.
• Anything about health, disability, learning needs or behaviour.
• Family details such as separations, money problems or court matters.
• Photos, voice recordings and scans of a child's work with a name on it.
• Bank details and login details.

When you need the real details in the final message, add them yourself after the tool has finished. It takes seconds.

Why the caution

The reason is what happens to a chat once you send it. Some tools may keep it and use it to improve their models, depending on the plan and your settings.

The Australian privacy regulator recommends, as best practice, that organisations avoid entering personal information into public chat tools. It singles out sensitive information as the most important to keep out¹.

It also says to read the terms. Find out whether the tool's maker can get at the data you type in or the text the tool produces¹.

That is guidance for organisations covered by Australian privacy law. Whether it covers a one-person tutor depends on your own situation, so check with the regulator.

Read the data settings of your tool

Each maker explains its data handling on its own help pages. Look for the part that says whether your chats are used to train models, and whether you can turn that off.

Two examples follow. Both pages were checked on 31 October 2026 and are likely to change²³.

• OpenAI says its individual plans may use your content for training. If you give a thumbs up or down,² "the entire conversation associated with that feedback may be used to train our models".
• Anthropic says³ "Your Incognito chats are not used to improve Claude, even if you have enabled Model Improvement in your Privacy Settings".

Notice what that tells you. Settings differ between tools, between plans and even between chat modes. A setting that protects one chat may not cover another. Check it before each new tool, and look again every few months.

Where to check the privacy rules

This site doesn't give legal advice, and it can't say whether your use is allowed. Go to the official body for your country.

• Australia: the Office of the Australian Information Commissioner (OAIC). Its page on commercially available AI products is a good place to begin.
• United Kingdom: the Information Commissioner's Office (ICO). It runs a section for⁴ "small-to-medium-sized enterprises, start-ups, sole traders, small charities, groups and clubs". It also publishes guidance on AI and data protection. A notice on that guidance says⁵ "Due to changes made by the Data (Use and Access) Act, this guidance is under review and may be subject to change." So check the date on whatever you read.
• United States: the Federal Trade Commission (FTC) is the usual starting point on privacy for businesses. Rules about children's data can vary by state and by who you work for, so look up what applies to you.
• Elsewhere: search for your country's data protection authority.

If you work through a school, an agency or a tutoring platform, read their rules too. They may ban AI tools for student data altogether.

Habits that help

• Keep names in your booking system or diary, not in chat history.
• Delete old chats you don't need.
• Use one tool you've read up on, rather than trying several.
• Tell parents in your terms if you use an AI tool for drafting, and that you don't put their child's details into it.
• Never ask a tool to judge a child's needs or write anything clinical. See what AI can and can't do for the jobs to keep.

When a parent asks, be able to say plainly what you do and don't paste. That one answer builds more trust than any policy page.