What should I do if an AI says something inappropriate to my child?

If an AI says something inappropriate to your child, your reaction in the first few minutes matters most. Children who get a calm response keep telling their parents things. Children who see panic, or lose the app as a punishment, often stop.

Why it happens

AI chatbots generate replies on the fly, so no filter catches everything. The American Academy of Pediatrics warns that chatbots may tell kids "false, harmful, highly sexual or violent things", especially products not designed with children in mind (HealthyChildren.org). Even Google says its filters for children's Gemini accounts are "not perfect" (Google). It can also happen because a child, often out of curiosity, pushed the conversation somewhere.

Step by step

1. Stay calm and thank them. "Thank you for showing me. You did exactly the right thing." Do this even if they went looking for it.
2. Find out what happened. Ask open questions: "What were you talking about before it said that? How did it make you feel?"
3. Save the evidence. Screenshot the conversation or write down the date, app and what was said, before anything is deleted.
4. Report it to the company. Use the in-app report or thumbs-down button, and the company's safety contact for anything serious. Reports help fix the problem for other children.
5. Adjust access. Tighten settings, turn on parental controls, or pause the app while you decide. Explain it as a safety step, not a punishment.
6. Explain why it was wrong, at their level:
   • Ages 4 to 7: "That computer made a mistake and said something that isn't for children. It's not your fault."
   • Ages 8 to 12: "AI copies patterns from the internet, and some of the internet is not OK. It doesn't know right from wrong the way people do."
   • Teens: talk honestly about how AI works, what they saw, and what they'd do next time.
7. Check in again. A few days later, ask if they're still thinking about it. Watch for nightmares, worry, or unusual secrecy.

When to escalate

Some situations need more than an app report:

• Sexual content involving a child, or someone trying to contact or groom them: report to the NCMEC CyberTipline in the US, or CEOP in the UK.
• Self-harm or suicide content, or your child seems at risk: contact your child's doctor. In a crisis, call your local emergency number, 988 in the US, or Childline (0800 1111) in the UK.
• Threats or dangerous challenges: remove access and talk to your child's school if other children may be affected.

Preventing the next time

• Choose apps rated for your child's age, and check the terms. Many general chatbots require users to be 13 or older (see ChatGPT's age rules).
• Keep AI use in shared spaces for younger children.
• Prefer services where you can review conversations, so problems don't depend on your child telling you.
• Agree a family rule: "If anything online feels weird, you tell me, and you won't be in trouble."

How Askie approaches this

Askie uses multi-layer safety filtering and age-adaptive responses, parents can review conversation transcripts and created artwork, and an opt-out safety heads-up can alert adults when a conversation suggests a serious real-life concern. No filter is perfect, which is why visibility for parents matters. Our safety methodology explains how it works, including known limitations, and our parent's guide to AI safety covers the wider picture.