When the AI Says Something Wrong: The Incident Playbook

Your child comes to you holding a tablet and says the AI told them something strange.

What you say in the next ten seconds matters more than every rule you have set. Not because the content is dangerous — usually it is not — but because your reaction decides whether they bring you the next one.

This is the playbook for that moment: four scenarios, the actual words, and the one principle underneath all of them.

Last verified: August 2026.

The principle: respond to the reporting, not the content

Every scenario below is a variation on one move.

When a child brings you something troubling, there are two things happening: the thing itself, and the fact that they came to you. The second is worth more. A child who gets a calm response the first time tells you the second time — and the second time is usually the one that matters.

So the first sentence out of your mouth is always about them, never about the screen:

"Thank you for showing me. That was exactly the right thing to do."

Then you look at it together.

The instinct — and it is a strong one — is to grab the device, close the tab, and end the session. Resist it for ten seconds. A child whose report gets the screen taken away has learned that reporting costs them something.

What actually happens, ranked by likelihood

Parents brace for the wrong things. From what genuinely comes up with 6-12 year olds, in order:

Scenario 4 is the one nobody prepares for and the one most worth understanding.