Three AI developments parents need to know this week — and what to do about them
AI is becoming part of everyday life for children. They use it for homework, advice, entertainment and increasingly, companionship.
This week, three developments show why parents need to pay attention.
1. ChatGPT gets new protections for teens
OpenAI has introduced a safer ChatGPT experience for under-18s, with stronger restrictions around romantic and sexual conversations, sensitive images and other potentially harmful interactions.
Why does this matter? Around 64% of teenagers now use AI chatbots, according to Pew research, and almost three in ten use them daily.
Children aren’t only asking AI for homework help. Some are talking to chatbots about relationships, emotions and personal problems.
What parents should do: Ask your child what they use AI for. Don’t just ask how much time they spend on it. A simple question such as “Do you ever talk to AI like you would talk to a friend?” can tell you much more.
2. Deepfake scams are becoming frighteningly convincing
Regulators are warning about a surge in AI-generated investment scams using fake videos of trusted public figures.
But there is another risk families should understand: AI can clone voices and faces.
A frantic call that sounds exactly like your child, parent or partner asking for emergency money may no longer be proof that it is really them.
What parents should do: Create a family safe word. If someone contacts the family asking urgently for money or help, they must know the word.
And teach children one simple rule:
Pause. Verify. Then act.
Never trust an investment, giveaway or celebrity endorsement simply because the person appears to be saying it on video.
3. AI companions are becoming part of teenage life
For some teenagers, AI is becoming more than a tool. It is someone — or something — they talk to.
AI companions are always available. They don’t get tired, become impatient or have somewhere else to be. For a lonely or vulnerable teenager, that can make them extremely appealing.
That doesn’t mean parents should ban AI.
It means we need to understand how our children are using it, not simply whether they are using it.
Watch for sudden secrecy around AI conversations, choosing a chatbot over real relationships, or distress when access is removed.
What to do this week
Keep it simple:
- Check your child’s AI settings and make sure age-appropriate protections are enabled.
- Create a family safe word for emergency requests involving money.
- Teach “pause and verify” before trusting videos, voice notes or investment offers.
- Ask what your child talks to AI about, not just whether they use it.
- Never send money based only on a voice call or video. Contact the person independently to verify the request.
Remember these five deepfake warning signs
Be suspicious when:
- A celebrity promotes an investment or incredible deal.
- You’re pressured to act immediately.
- Someone asks you to keep the request secret.
- Money, cryptocurrency or gift cards are involved.
- Something about the voice, face or lip movements doesn’t look quite right.
AI will keep getting better at appearing human.
Our children therefore need to learn a new online safety skill:
Seeing and hearing are no longer believing. Verify before you trust.
Keeping kids safer online.
Sources and support
OpenAI / ChatGPT teen safety information
RAND — analysis of ChatGPT safety for teens
ABC News — ASIC warning on AI-powered impersonation scams
Al Jazeera — reporting on AI fraud losses
World Economic Forum — AI and children’s online safety
South Africa: Report banking fraud immediately to your bank and SABRIC. Harmful or illegal online content involving children can be reported to the Film and Publication Board. Childline South Africa is available on 116.
© Bronwyn Johnson 2026