Field notes

The parent guide to safe prompting

By Ashley · Build Metacognition · · 3 min read

SafetyPrivacyFor parents
The short version

Safe prompting is not prompt engineering. It's three habits: keep private information out, ask AI to explain rather than answer, and verify anything that sounds like a fact. Your child does not need to become a prompt engineer.

Safe prompting is not about writing fancy prompts.

It is about helping kids use AI without giving away private information, copying answers, or trusting everything the tool says.

That is the parent version.

Your child does not need to become a prompt engineer. They need to learn how to ask for help in a way that keeps them safe, honest, and thinking.

Rule 1: do not share private information

Start here.

Students should not put private information into AI tools.

That includes:

The simple rule:

If you would not want it saved, shared, or seen by someone else, do not put it into AI.

Rule 2: ask for help, not replacement

A safe learning prompt keeps the student active.

Unsafe:

"Write my essay."

Safer:

"Ask me three questions to help me plan my essay. Do not write it for me."

Unsafe:

"Solve this math homework."

Safer:

"Give me one hint for the first step. Do not give me the answer."

Unsafe:

"Make this sound smart."

Safer:

"Tell me one sentence that is unclear and why."

This is the difference between support and shortcut.

Rule 3: make AI explain, then check understanding

A good prompt asks AI to teach, not perform.

Try:

"Explain this concept in simple words. Then ask me one question to see if I understood."

or:

"Give me an example, then ask me to make my own."

or:

"Quiz me one question at a time. Wait for my answer before explaining."

These prompts make the student respond. That matters.

Learning requires effort. AI should not remove all effort. It should aim the effort better.

Rule 4: ask what could be wrong

Students need to learn that AI can be confidently wrong.

A strong prompt:

"What parts of your answer should I verify?"

Another:

"List three claims that need a source."

Another:

"What might be missing from this explanation?"

This teaches skepticism without cynicism.

The message is not "never trust AI." The message is "check before believing."

Rule 5: keep the student's voice

AI can easily rewrite a student's work until it no longer sounds like them.

That is risky.

Instead of asking AI to rewrite, students can ask for feedback:

"Tell me what is confusing, but do not rewrite it."

"Give me one suggestion to make this clearer."

"Point out where I need more detail."

The student should make the changes.

That keeps authorship where it belongs.

Rule 6: follow the teacher's rules

Every class may handle AI differently.

Some teachers allow brainstorming. Some allow grammar feedback. Some allow study help. Some do not allow AI on specific assignments because they need to see what students can do independently.

Students need to learn context.

A safe prompt does not make AI use automatically allowed. The assignment rules still matter.

When in doubt, ask first.

A safe prompt formula

Use this structure:

  1. I am learning...
  2. I already tried...
  3. Help me by...
  4. Do not...
  5. Ask me to respond.

Example:

"I am learning how to identify theme in a story. I already read the chapter and wrote two possible themes. Help me compare them. Do not write my answer. Ask me one question at a time."

That is a strong student prompt.

It gives context. It keeps the student active. It blocks answer-copying.

A parent script

Try saying:

"You can use AI to explain, quiz, and give feedback. You cannot use it to do your work for you. And you need to check anything factual."

That is enough to start.

The next step

Keep the thinking with the student.

Build Metacognition teaches students safe prompting, honest use, fact-checking, and project-based AI literacy.

Join the founding waitlist

Sources