The old advice was to look for strange spelling and obviously edited photos. AI is getting rid of the obvious bit.
I used to think fake things online would look fake.
The hands would have seven fingers. The voice would sound like a robot inside a tin. The message would say something like "HELLO DEAR FRIEND" and ask for your password before getting your name right.
That was convenient.
A new survey found that 44 per cent of New Zealand parents were not confident their child could tell real content from something made by AI. Three quarters were worried AI could be used to create fake pictures or conversations involving their child.
This is not really a story about whether kids are smart enough. Very smart adults get fooled too. It is a story about the evidence changing.
A photograph used to feel like proof. A recording used to feel like proof. Now an ordinary person can make a convincing fake picture, voice or message in minutes.
Gaming matters here as well. Among New Zealand parents in the survey whose children used games with voice or text chat, 49 per cent said their child had chatted with a stranger.
I do not think the answer is to panic every time somebody online says hello. It is to build a new habit before trusting or reacting:
Pause. Is this trying to make me scared, excited or rushed?
Check. Can I confirm it somewhere else, especially from the person or organisation it claims to be from?
Tell. If a message becomes secretive, threatening, personal or asks me to move to another app, show a trusted adult. Keeping it secret protects the sender, not you.
There is another rule adults need to learn too: ask children before posting their photos publicly. A picture can be copied, changed and reused in ways the person in it never agreed to.
The internet used to tell us, "Do not believe everything you read."
Now we need the upgraded version: do not automatically believe everything you can see or hear either.
That is annoying. It is also learnable.