This week’s online safety news brought three warnings every parent should know about, and simple steps you can take at home today.
AI companions, deepfakes and increasingly powerful AI tools are creating new risks for children. This week, research suggested that one in three children would rather talk to an AI companion than a real person, while child-safety experts continue to warn about companion chatbots and AI-generated explicit images. Parents do not need to understand every new technology. But we do need to understand the risks and set sensible boundaries.
Here is what matters this week.
1. AI companions are becoming real companions
AI companions are chatbots designed to feel like friends, romantic partners or confidants. They remember conversations, respond instantly and can provide seemingly endless attention.
A recent report from the Information Technology and Innovation Foundation (ITIF) found that one in three children prefer talking to an AI companion rather than a real person.
That matters.
Earlier this year, Common Sense Media, working with Stanford researchers, rated AI companion platforms unacceptable for anyone under 18.
US regulators are also investigating how companion chatbots affect children and teenagers.
The concern is not simply that children use AI. It is that some may begin relying on a machine for friendship, reassurance or emotional support that should come from trusted people.
2. Deepfakes are getting easier to create
A deepfake is an AI-generated or altered image, video or audio recording that looks real.
For families, one of the most disturbing developments is so-called “nudify” technology, which can turn an ordinary photograph into a fake explicit image.
French authorities are investigating sexually explicit AI-generated images, including images involving teenagers, reportedly created using Grok.
A single photograph from social media can be enough.
For a teenager whose fake image is circulated at school or online, the consequences can be devastating.
There is some progress. New European Union AI transparency requirements are coming into force, including requirements around identifying AI-generated and manipulated content.
But technology and regulation alone will not solve the problem.
3. Children can access powerful AI tools easily
Children no longer need technical skills to use sophisticated AI.
They can download an app, start chatting to an AI companion or upload a photograph to an AI tool within minutes.
And many digital products are deliberately designed to keep people engaged.
Children are particularly vulnerable because their judgement, emotional regulation and understanding of relationships are still developing.
A chatbot that is always available, always interested and rarely disagrees can become surprisingly compelling.
That makes conversations at home increasingly important.
Five things parents can do this week
1. Ask about AI companions.
Find out whether your child uses Character.AI, Replika or similar services. Ask with curiosity rather than immediately banning or criticising them.
2. Explain what the chatbot really is.
Make sure your child understands that an AI companion is software, not a friend, therapist or trusted adult.
3. Check their apps.
Review installed apps, age ratings and parental controls. Android families can use Family Link, while Apple devices include Screen Time controls.
4. Talk about deepfakes before something happens.
Explain that someone can create a fake explicit image from an ordinary photograph. Most importantly, make sure your child knows they should tell you immediately if it happens.
5. Review their public photos.
Check privacy settings and who can view or download your child’s photographs. Fewer publicly accessible images mean fewer opportunities for misuse.
What if something has already happened?
If someone creates or shares a fake explicit image of your child, keep the evidence and act quickly.
In South Africa:
- Report child sexual abuse material to the Film and Publication Board.
- Report suspected criminal activity to SAPS.
- Report the image directly to the platform and request removal.
- If your child needs emotional support, Childline South Africa provides free assistance on 116.
Creating or distributing sexual images involving children is a serious matter. Do not dismiss a fake image because “it isn’t real.” The harm to the child can be very real.
You don’t need to keep up with every new AI tool
AI will continue changing faster than most parents can follow.
The goal isn’t to know every app.
It is to create a relationship in which your child will tell you when something feels uncomfortable, frightening or wrong.
Start with one question this week:
“Have you ever chatted to an AI like it was a real person?”
The answer may tell you more than checking their phone ever will.
Click Safe Online is here to make new online risks easier to understand and give families practical ways to respond.
Keeping kids safer online.
Sources and resources
Research and reporting
- Information Technology and Innovation Foundation (ITIF), Better Chatbot Safety for Children Requires Smarter Policy, 10 August 2026.
- Common Sense Media and Stanford research on AI companion safety for teenagers.
- Federal Trade Commission investigation into AI companion chatbots.
- European Commission guidance and AI Act transparency requirements.
- Reporting on AI-generated explicit deepfakes involving women and teenagers.
© Bronwyn Johnson 2026