Trust & safety — how Pace & Race handles your child and AI
Pace & Race Adventures is built for kids ages 7–13. This page says exactly what our AI does with your child, what it never does, and what we never do with their words.
Your child never hands us anything. You hold the keys.
Kids can start playing right away — no sign-up wall. No game asks for a name, email, school, or anything that identifies your child. The one exception: Storybook Quest asks for a nickname and an age so the hero can belong to somebody — and both can be made up.
Parent accounts are separate. The Family Hub, where you see your child's progress and creations, only opens with a grown-up's sign-in.
And a safety check runs before your child's words ever reach the picture maker. It's designed to block mentions of real people — names, family members, friends, teachers — and personal details. It's deliberately strict, and it sometimes flags a harmless sentence; when it does, the game doesn't scold. A character warmly steers them back, and their words stay intact so they can edit and try again.
We don't keep AI chats. What your child makes stays in your family.
Question Quest conversations live in memory while your child is chatting and aren't saved to their account. When the chat ends, there's no stored copy.
What your child creates — their stories, their pictures, their inventions — belongs to your family. We don't sell it, we don't share it, and we don't use it to train AI models.
The AI's instructions and access keys stay on our server and never reach the browser.
The AI is a helper, not a friend.
It doesn't pretend to be a buddy, and it doesn't remember your child between visits — every debate, every chat, starts fresh. It never asks for personal details. In every world, it has one job: help with the thing your child is making, or the question they brought.
In every world, your child is the author. The AI is the assistant.
This is the rule across the whole platform — AI as collaborator, not replacement. It maps to the two frameworks schools use for AI literacy: AI4K12's Five Big Ideas (the K-12 AI guidelines from AAAI and CSTA — Perception; Representation & Reasoning; Learning; Natural Interaction; Societal Impact) and the OECD's AI Literacy Framework (Engage with AI; Create with AI; Manage AI; Shape AI — written for parents and caregivers too).
Canvas Cove
Better words make better pictures. Race types “DOG” and gets a plain brown dog. Pace asks one question, your child answers in their own words, and the AI paints what they described — one new skill per island. After every picture, Pace asks “What did your new words change?” and your child answers in their own words, saved beside the picture in My Stuff.
Why it matters: Learning to tell an AI exactly what you mean is becoming a foundational skill, and kids learn it fastest by watching the picture change when their words change. In a study of 10 children ages 7–12 using picture-making AI, kids showed high engagement in exactly this loop: trying new words, watching the image change, revising again (Han et al., 2023). Classroom researchers have built this into a “prompt-to-pedagogy” approach where children suggest keywords, evaluate the images, revise, and explain why a prompt should improve — the revision itself becomes the learning (Wichitchokthananon). The same research warns of “prompt dependence” — kids leaning on the AI instead of thinking — which is why Canvas Cove only rewards the child's own typed words: tapping a suggestion earns nothing.
Frameworks: AI4K12 Big Idea 4 (Natural Interaction) and Big Idea 2 (Representation & Reasoning) • OECD Create with AI — directing generative AI to elicit feedback, refine results, and support reflection.
Revision World
Make a good idea better. Your child builds one picture from their answers, then revises it one small change at a time — a describing word, a stronger word, a new style — and watches the picture change with their words. Pace's closing line: “every change was your idea. that's what revising is.”
Why it matters: Revision World trains the judgment muscle: not just getting an image, but deciding whether it's right and knowing which word to change. A 2026 research review found that the children who got the most from generative AI were the ones who learned to “craft iterative prompts and critically evaluate AI-generated outputs” — while younger children needed structured, visual routes to get there (Frontiers in Psychology, 2026). That's what this world is: a structured route where each island isolates one decision — one word, one style, one change — so the child practices evaluating the AI's work instead of just accepting it.
Frameworks: AI4K12 Big Idea 4 (Natural Interaction) • OECD Create with AI — refining results and supporting reflection.
Storybook Quest
Write a story that is all yours. Your child invents the hero, the place, and the problem, then writes each page with the AI's help — the AI illustrates the pages and polishes the prose, but it never invents the story. “Write my page!” stays locked until the child fills every story beat.
Why it matters: The fear with AI and creativity is that the machine does the creating. The research says the outcome depends on who stays in charge. In a study of 12 kids ages 7–13 directing AI art tools, children developed real creative self-efficacy — the belief they could boss the AI around — pushing back with things like “I don't think this really expresses what I'm saying” (Newman et al., CHI 2024). But a second study of 28 children ages 5–11 found the flip side: when roles were unclear, the AI dominated and children's agency shrank; kids themselves envisioned the ideal AI as an “older sibling” — scaffolding, while allowing playful disagreement (Davis & Yip, CHI 2026). Storybook Quest is built for the first outcome: the AI's suggestions appear only if the child taps for one, they're editable (“make it yours”), and every plot decision stays with the child.
Frameworks: AI4K12 Big Ideas 2 (Representation & Reasoning), 4 (Natural Interaction), and 5 (Societal Impact — thinking about who made what) • OECD Create with AI — using AI as a creative partner while preserving human agency and originality.
Fix-It Factory
Try. Tinker. Discover. Your child names a real problem, invents a fix, explains how it works, and earns a patent certificate for it. Race only offers ideas if the child asks.
Why it matters: Learning by naming problems and designing solutions isn't just fun — it's one of the best-studied ways kids build lasting skills. A 2024 review in the International Journal of Technology and Design Education looked at three controlled studies (Gözen, 2015; Yalçin & Erden, 2021; Yalçin, 2022b) and found that STEM activities taught through design-thinking methods produced significant gains in children's learning, innovation, life and career skills, and technology skills — gains that were still there on follow-up tests, while control groups didn't improve. Fix-It Factory is that loop in game form: spot a problem, invent the fix, explain how it works, own it.
Frameworks: AI4K12 Big Idea 5 (Societal Impact — inventing technology to help people) • OECD Create with AI and Shape AI — proposing how systems should serve human benefit.
Judge Pace
Defend your idea to Pace. Your child picks an opinion and defends it while Pace asks follow-up questions. There are no wrong opinions: Pace challenges the thinking, never the child, and every round ends “APPROVED!” for the effort.
Why it matters: Defending an idea out loud is how children learn to reason. In a UK randomized trial of about 4,800 nine- and ten-year-olds across 80 schools, talk-intensive lessons — where questions invited more than recall, and answers were justified, followed up, and built upon — produced two months' extra progress in English and science compared to normal lessons (Education Endowment Foundation, 2017). Judge Pace bottles that dynamic: Pace's questions (“What would you do for kids who need that sleep or all their lunch time?”) force the child to consider the other side, and a “How I thought of that” button explains why Pace asked — so the child sees reasoning modeled, not just demanded.
Frameworks: AI4K12 Big Idea 2 (Representation & Reasoning) • OECD Engage with AI — interacting with conversational AI and evaluating how it responds.
Question Quest
A Socratic tutor for the question stuck in their head. It asks a question back first, gives one small hint and waits, then checks your child's reasoning gently — pointing at the wobbly part and letting them fix it themselves. It's instructed, in rules the browser never sees, to never give the direct answer and never write your child's work for them. Stuck after three exchanges? They can ask for one bigger hint. Still not the answer — and no rewards for pressing it.
Why it matters: A 2025 Wharton study published in PNAS found that students given an AI that dispensed answers scored 17% lower once the AI was taken away — the AI had done their thinking for them. Students given an AI that gave hints instead largely avoided the harm. We built the hint-giving kind. (Bastani et al., PNAS, 2025.)
Frameworks: AI4K12 Big Idea 2 (Representation & Reasoning) • OECD Engage with AI.
No tricks that keep kids hooked.
No streaks. No hearts or lives. No leaderboards. No timers or countdowns. No visible scores. Nothing is ever taken away, and retry screens are warm — never red, never punishing. The only rewards are Shells, and they're earned for your child's own words. Never for speed.
On the new school AI rules
Schools are writing new AI rules, and we'll keep this page current as they land. We don't claim certificates or seals we haven't earned — we publish what the product actually does, and you can check every claim on this page against the game itself.
Here are the rules on the way, what they ask for, and where each world fits.
U.S. executive order on AI education (April 2025)
What it asks for: A national push to teach AI literacy from the earliest grades, so children learn how AI works and how to use it well, not just how to tap it.
Where we fit: Every world teaches one AI skill by doing it. Canvas Cove and Revision World teach directing and refining an image maker; Question Quest teaches asking AI for hints, not answers.
Children's privacy rule update (COPPA, compliance from April 2026)
What it asks for: Stricter limits on collecting children's information, separate parent consent before a child's data is shared or used for things like AI training, and clear data-keeping limits.
Where we fit: No game asks for a child's name, email or school. Question Quest chats aren't saved. Creations are never sold, shared or used to train AI. Storybook Quest's nickname and age can be made up.
Illinois Public Act 104-0399
What it asks for: State guidance for schools covering basic AI ideas such as machine learning and computer vision, keeping human relationships at the heart of teaching, and protecting student data.
Where we fit: Chip shows how AI learns from the examples it's given. Judge Pace keeps the thinking with the child: Pace asks, the child reasons. A grown-up's sign-in guards the Family Hub.
Maryland Artificial Intelligence Ready Schools Act (2026)
What it asks for: Every county needs an AI policy and an AI coordinator, and schools must pick AI tools that match the state's guidelines, with AI literacy built into what students learn.
Where we fit: This page is written so a coordinator can check us against those guidelines in plain words. Each world's grown-ups note names the OECD AILit domains and the AI4K12 draft guidelines it covers.
Idaho, Arizona, Hawaii and other states (bills in progress)
What it asks for: Proposals ask for AI policies in every district, AI lessons at set grades, prompt-writing skills and the ethics of using AI.
Where we fit: Storybook Quest and Fix-It Factory teach the ethics side: the child is the author and inventor, and the AI only helps. Canvas Cove only rewards the child's own typed words.
None of this makes us approved by any state or agency. Each district picks its own tools; this is a map to help you and your child's school decide.
Questions about any of this?
Email aliyah@paceandrace.ai and we'll answer plainly.