Three Age-Specific Verification Games for Kids 7 to 16
Three games that teach kids to verify AI outputs — one each for ages 7-9, 10-12, and 13-16. No curriculum, no tech required. Each takes under 20 minutes and builds the habit that matters.
Three Age-Specific Verification Games for Kids 7 to 16
Your seven-year-old asks Google Assistant what the biggest animal in the world is and accepts the answer without question. Your twelve-year-old uses ChatGPT to research a history project and copies three facts directly into her notes. Your fifteen-year-old writes an essay using AI as a starting point and submits it having never checked whether the quotes attributed to real people are real.
These are three different problems, but they share the same root: none of these kids has been given a reason to pause before accepting what a machine told them. And the reason they haven't been given that pause isn't negligence — it's that the habit of verification has to be taught deliberately, and it has to be taught in a way that makes sense for where they are developmentally.
The games in this post are built around one core idea we've developed across this series: the "How do you know?" habit isn't a single skill. It shows up differently at seven than at twelve than at fifteen, and activities that work well for one age band often fall flat for another. A teenager finds the games designed for younger kids condescending. A second-grader gets lost in the nuance designed for middle schoolers. Each of the three activities below is designed to land where the child actually is.
You don't need any technology to run any of these. You don't need to understand AI yourself. You just need fifteen or twenty minutes and a willingness to be wrong alongside your kid.
Ages 7 to 9: The Robot Reporter Game
Children at this age are concrete thinkers. Abstract reasoning about why sources might be unreliable is too far ahead of where their development is. What works instead is giving the AI a persona that makes skepticism feel natural, pairing it with physical checking, and keeping the stakes low enough that being wrong is funny rather than threatening.
Setup: Tell your child that you're going to interview a robot reporter who knows a lot of things but sometimes gets facts mixed up. Their job is to be the fact-checker — the person who makes sure the robot's stories are true before they get published.
How to play: Ask the AI (ChatGPT, Siri, Google, whatever your household uses) a question about animals, space, or a topic your child finds genuinely interesting. Something specific enough to be verifiable: "How fast can a cheetah run?" or "How far away is the moon?" or "How long do elephants live?" Write down the answer the AI gives.
Now the fact-checking begins. Pull out a children's encyclopedia, a book from the shelf, or a trusted reference website you navigate to together — not another AI search. Look up the same fact. Compare what you find to what the robot reporter said.
If the AI was right, your child gives it a thumbs up and the story is approved for publication. If it was wrong or slightly off, your child marks it as a correction needed and explains what the real answer is.
What you're building: The habit of checking before accepting, practiced in a context where checking feels like the job rather than an act of distrust. The persona does the work here — "fact-checking the robot reporter" is a role with clear authority, and children at this age respond to roles. You are not asking them to doubt the technology. You are asking them to do a satisfying job.
Run this two or three times across different topics and you'll notice your child starting to ask the verification question without prompting. That's the habit forming.
Ages 10 to 12: The Citation Hunt
Children in this range can handle more complexity. They're starting to do research for school, which means they're encountering AI outputs as potential source material rather than just trivia answers. The critical thinking skill they need at this stage isn't just "is this fact right?" — it's "where did this claim come from, and how would I know?"
AI language models frequently cite sources that don't exist. They attribute quotes to real people who never said them. They reference studies with plausible-sounding titles that have no corresponding paper. This is one of the most common and consequential failure modes, and it's the one most invisible to a child who has learned to accept cited information as verified information.
Setup: This one works best when your child is working on an actual assignment, but it can also be done as a standalone activity. Ask the AI to give you information on any substantive topic — a historical event, a scientific concept, a notable person's contributions — and specifically ask it to include sources.
How to play: Take one of the sources the AI provides and hunt for it. Not to see if it agrees with the AI's claim, but to confirm it actually exists. Google the exact title. Look for it on the publisher's website. Try to find the page number, the author's institution, the publication date.
When you find the source is real, follow up: does the source actually say what the AI claimed it said? Pull the relevant section. Read it together. Compare it to what the AI wrote.
When the source doesn't exist — and this will happen — that's the moment that sticks. The AI didn't say "I'm not sure." It didn't flag uncertainty. It produced something that looked like a citation and wasn't. Having that experience once, at age eleven, creates a reference point that lasts for years.
What you're building: A specific and transferable skepticism about citations. The rule that emerges naturally from this game is: a citation isn't verification, it's an invitation to verify. That rule applies to every subject and every tool they'll use for the rest of their education.
One hunt per week is enough. The goal isn't to spend hours checking every source — it's to develop the reflex to check at least one, every time.
Ages 13 to 16: The Confidence Calibration Challenge
Teenagers are capable of genuine metacognition. They can think about their own thinking, evaluate their own certainty, and update their views when given evidence. The skill gap at this age isn't abstract reasoning — it's that they often don't apply that capacity to AI outputs specifically. AI sounds authoritative. It doesn't hedge the way a person hedges. It produces confident prose even on topics where confidence isn't warranted, and teenagers who have developed good critical thinking in other contexts sometimes leave that thinking at the door when they're using AI as a tool.
This game targets that gap directly.
Setup: Ask the AI a series of five questions across different types of topics. Mix in at least one question about a recent event (anything in the last six months), one that requires a precise numerical fact, one about a public figure's specific statements or actions, and one that's well-established and stable. Write down the AI's answers.
How to play: Before looking anything up, your teenager rates their confidence in each answer on a scale of one to five — one being "I suspect this is wrong" and five being "I'd be surprised if this were inaccurate." Then they verify all five, using primary sources when possible.
The debrief is the game. Where were they right to be confident? Where were they confident and wrong? Where were they skeptical but the AI was actually right? Most importantly: what pattern do they notice? AI tends to be more reliable on stable, well-documented topics and less reliable on recent events, specific quotes, and precise figures. Does their calibration match that pattern?
Run this a few times over a few weeks and your teenager will start building an internal model of AI reliability — not "AI is right" or "AI is wrong" but a more nuanced sense of what types of questions are higher-risk and why.
What you're building: Calibrated skepticism. The goal isn't maximum distrust of AI — that's as unhelpful as maximum trust. The goal is accurate confidence: knowing when to use AI outputs directly, when to spot-check, and when to verify thoroughly before relying on what it produced. This is the skill that transfers directly to professional use, because every adult who uses AI well has developed some version of this internal calibration model.
A Note on All Three Ages
None of these games requires your child to stop using AI. None of them are designed to make AI feel dangerous or untrustworthy as a category. The framing throughout is empowerment: you are the one in charge here, and being in charge means knowing when to check.
As we discussed in our post on the two types of critical thinking, the skill worth building isn't thoughtful usage — it's the habit of directing critical thinking toward the AI rather than only applying it to the assignment downstream. These games are the on-ramp to that habit, scaled for where each age group is developmentally.
The Spot the Lie game we described earlier in this series is still a strong option across all three age bands for a group or family setting. The three games above are designed for one-on-one or small-group work where you can have a real-time conversation about what the verification process is revealing.
And the foundational question from our post on the "How do you know?" habit — asking it consistently, across contexts, until it becomes automatic — is the thread that connects all of this. The games make verification concrete. The habit makes it permanent.
Next week we're covering a harder story: what happens when adults who should have known better trusted AI completely, and what that tells us about what we're asking our kids to navigate without any of the guardrails adults take for granted. The case studies are specific, the stakes were real, and the lessons transfer directly to skills kids can start building now.
If you want to be notified when that post goes live, and when we open access to the full AI literacy curriculum we're building for K-12 families launching August 2026, join the waitlist.
Orbrya builds AI literacy curriculum for K-12 families — homeschool and supplemental — focused on the verification and evaluation skills that separate effective AI use from dependent AI use. Curriculum launches August 2026.