You cannot pre-debunk the internet. You can build the kid who checks. The whole arc of this conversation is one filter, installed early and upgraded yearly: who made this, why did they make it, is it true. By sixteen it should run automatically, which means at four it starts as a game.
What they're really learning: the real-versus-pretend line, and the first shocking truth of media literacy: everything on a screen was made by a person who chose what to show.
Say: "Is that real or pretend?" about everything, playfully: cartoons, commercials, dinosaur documentaries. Then the upgrade question: "Somebody MADE that show. They picked what we see. I wonder why they picked that?"
And the first ad lesson: "That's a commercial. Its whole job is making us want things. Silly commercial, we see what you're doing."
Skip: expecting the line to be crisp; four-year-olds think pretend things live somewhere real. That's fine. You're planting the question, not harvesting the skill.
What they're really ready for: running the filter themselves: who made this, why did they make it, is it true.
Say: "The internet is full of true things, mistaken things, and trick things, and they all look the same at first. So we check like detectives: who made this? Why, are they selling, joking, or informing? And how could we find out if it's true?" Then actually check one together, weekly, on something small and fun. Is that animal fact real? Detective time.
The feeling flag, installed early: "When something makes you feel a big feeling FAST, that's the moment to slow down. Big fast feelings are how tricks work." Word-for-word: the is-the-news-real script.
Skip: making it about politics. At eight the reps come from dinosaur facts and viral animal videos. Neutral reps build the muscle that handles loaded content later.
What they're really facing: AI-generated content indistinguishable from real, algorithms that reward outrage, and a feed that arrives via friends, pre-trusted.
Say: "New rule for the AI era: seeing is no longer believing. Photos, voices, and videos can all be generated. So the question shifts from 'does it look real' to 'who benefits if I believe it.' And know how the feed works: apps make money on your attention, and anger holds attention best, so the angriest version of everything rises. The feed isn't showing you the world. It's showing you what keeps you scrolling."
Give them the verification moves: check the date, search the claim plus the word "fact check," look for the original source, reverse-search the image. Four moves, two minutes, most fakes dead on arrival.
Skip: total cynicism as the lesson. "Trust nothing" is what conspiracy funnels teach; it leaves a vacuum they fill. Teach "trust is earned by sources that show their work and fix mistakes out loud."
What they're really navigating: misinformation as a social problem: the relative who shares everything, the friend going down a rabbit hole, their own confirmation bias.
Say: "Hardest level of media literacy: catching it in people you love, and in yourself. Nobody was ever fact-checked out of a belief at Thanksgiving; people argue their way IN deeper. What works is staying connected and asking real questions: where'd you see that, what would change your mind? Same standard for us: what would change ours?"
And the self-audit habit: "The most dangerous misinformation is the kind you already agree with, because it doesn't trip your alarm. When something flatters your side perfectly, that's exactly when to run the three questions." The Values Drift Test makes this concrete, for them and for you.
Skip: deputizing them to fix the relative. Model the bridge instead: warm, curious, unmoved. If someone they love is sliding somewhere darker, the Decoder and the Map show the terrain, and Parents for Peace (1-844-A-BELONG) exists for families in real drift.
Who made this, why, is it true, upgraded yearly from cartoons to deepfakes to their own bias. The filter is the inheritance. Install it early enough and no algorithm gets to be their only teacher.