The feed is built
to fool them.
Teach them to see it.
AssumeZero trains children to question what they're shown — AI fakes, clickbait, manufactured outrage — using inoculation, a method shown to build real, lasting resistance to misinformation. The difference from everything else on offer: it's built in the format children already live in, so a proven method actually reaches them. v0.3 is a working prototype of that loop.
The skill is a snap judgment. So is the game.
Spotting manipulation in the wild isn't an essay — it's a two-second instinct, then a check. v0.2 proved the world was buildable, but the lesson arrived after too much reading, and read-not-played is exactly the failure mode of the provision schools already have. A child who has to do homework to learn this won't.
So v0.3 rebuilds the engine around the real skill. The core is a feed where the child makes a fast, visual call — real, or trying to fool me? — and only then does TruthNet reveal the trick, name the tactic, and reward the catch. The format that manipulates them becomes the format that trains them against it.
The same warm world. The same TruthNet lesson. A loop built around how the skill actually works.
Spot. Reveal. Repeat.
One judgment, dressed a hundred ways. The disguise changes; the question never does.
The feed
A card appears — a post, a photo, a clip. Real, or made to fool you? A two-second gut call. Swipe or tap. No wall of text, no homework face.
TruthNet cuts in
After the swipe, TruthNet names the tell — the seventh leg, the red arrow pointing at nothing, the feeling that was engineered. The "ohh" moment is the learning.
Tokens & boosters
Catches earn Truth Tokens. Spotting fades without practice — so the game asks you back tomorrow for a booster. The replay loop is the lesson's delivery schedule.
And then — make the lie. In Creator Mode the child stops spotting and starts building: pick a real photo, bolt on a panic caption, add the arrow, watch the likes inflate. Becoming the manipulator — briefly, safely, white-hat only — is the highest-potency version of the lesson, and the research agrees.
The lesson grows with the child.
The same loop carries a different question at each Key Stage — because what kind of lie a child can even perceive changes as they grow. It future-proofs the game: even when AI fakery becomes flawless, "who benefits from me feeling this?" never stops working.
5–7
7–11
11–14
This isn't a hunch about gamification.
It's inoculation theory — "prebunking." Expose a child to a weakened, controlled dose of a manipulation tactic and they build resistance to the real thing. It's been demonstrated in nearly this exact format: Cambridge's Bad News game conferred measurable resistance to misinformation across a ~15,000-person evaluation, holding regardless of age, education or politics — and the strongest effect came from letting players build the fakes, not just spot them. The EU Commission has called gamified prebunking one of the most sustainable paths to combating fake news.
The honest caveat we keep on the record: the effects are real but modest. We shift the odds. We don't immunise.
of online 13–17s shown a celebrity product post failed to confidently identify it as influencer marketing.
of online 8–17s now use AI tools — often for learning and schoolwork.
of 8–17s recall an online-safety lesson at school — almost all of it read, not played.
The Online Safety Act has created statutory pull toward exactly this kind of provision — and there still isn't much children will voluntarily engage with. That's the gap.
The village is the world. The feed is the engine.
v0.3 doesn't retire Little Fibbing — it gives the village a job it's good at. The warm world and the sharp TruthNet register were always "the same mechanic in two registers." Now the digital register is the loop children play, and the village is the story that connects the rounds and answers why — delivered in small, voiced doses, never as a gate in front of the fun.
Carried over from v0.2
- The Little Fibbing world & cast — narrative frame
- TruthNet as the resolution / reveal mechanic
- The dual warm / digital visual architecture
- Uncertainty-first: "I can't tell yet," not false certainty
- PSHE / RSHE alignment, KS-tiered
New in v0.3
- Swipe-to-judge feed as the core loop
- Provenance → framing → feeling spine across KS1–KS3+
- Truth Tokens + booster return mechanic
- Creator Mode — build the lie, white-hat
- Validated by playtest, not just buildable
Retired: the headline slider / Exagga-rometer as a primary mechanic — it leaned on words and dragged the pace. Its job (strip the hype, find what's left) now lives inside the reveal.
Where we are. What's next.
Complete — v0.2 POC (web)
Walkable village, NPC dialogue, the LIKELY / CAN'T TELL / UNLIKELY runner, three-state ad break, TruthNet cards. Proved the concept was buildable without a studio.
Complete — v0.3 core loop (web prototype)
Working swipe-to-judge feed across the spine — provenance (real vs AI), framing (clickbait), and a system-level trap; TruthNet reveals; Truth Tokens; a Kling-generated behind-the-scenes sting; Creator Mode; and the booster end-screen. Real images, real loop — a short, deliberate proof slice.
Complete — playtest, first round
Early testing with KS1 and KS2 children showed strong voluntary engagement, including repeat plays of the placeholder-art build — evidence the loop itself is the draw. Engagement is the delivery mechanism, not the goal; the next rounds test what matters most: whether the lesson sticks.
In progress — content volume
The loop holds attention; the deck is short. Building a repeatable card pipeline (GPT-image + Kling) so content scales without becoming a bespoke art project per card.
Upcoming — school pilot conversations
Seeking 2–3 UK primaries for endorsement and a home-play distribution pilot. No procurement — heads look at something and tell us honestly what they think.
Upcoming — designer engagement
Game-native visual direction for the feed, cards and TruthNet reveals, plus a repeatable content-art system. The interim build is house-style; the next pass needs a children's illustrator.
Educators. Designers. Funding.
Educators
KS1–KS3 teacher, PSHE lead or head? I want listening conversations — 30 minutes, no commitment — and 1–2 schools willing to endorse a pilot.
Get in touch →Designers
Game-native art for the feed and TruthNet reveals, plus a repeatable card-content system. Children's illustrator, UK preferred. Brief available.
See the brief →Funding
Pre-seed. Individuals or small funds who get the edtech and consumer opportunity in children's media literacy. The gap is real, documented, and now playtested.
Let's talk →The narrative world still exists. Right now.
Before the feed, there was the village. The v0.2 proof-of-concept — a walkable Little Fibbing, NPC dialogue, the headline runner and TruthNet cards — still runs in a browser. It proved the concept was buildable without a studio, and it remains the narrative world that v0.3's feed plugs into.
Little Fibbing — The Pothole Incident
Walk the village, follow the rumours, judge the headlines. A short, deliberately rough slice — the first thing AssumeZero ever shipped to a browser.
Play the v0.2 demo →David · Founder, Arcturus Digital Consulting
I'm a product manager and app developer in the UK. AssumeZero grew out of a question I couldn't drop: if the information environment is this demanding for children, why does so much of what we hand them still assume the internet looks the way it did fifteen years ago?
The aim is narrow and durable — teach children to question what they're shown, with a method that has evidence behind it, in a form they'll actually choose. v0.3 is the build that puts that skill at the centre of the loop.
Previous work includes Sprocket, STEa, and a tiered portfolio of consumer and B2B products under ArcturusDC.