Every story you read on The Scoop passes through several checks before it reaches a kid’s screen. Here’s exactly what those checks do, what they block, and how we keep a record of every decision.
Last updated: July 2026
The shape of a Scoop story
We aim for stories that a curious 6–9 year old can enjoy on their own, and that a parent would feel good about over the dinner table: real discoveries, helpers and heroes, sports victories, animals, space, climate progress, and community wins.
Our standard is affirmative: a story isn’t published because nothing bad was found in it — it’s published because its subject belongs to the kid world in the first place. When we’re not sure, we don’t run it. There’s always more news; a parent’s trust doesn’t come back as easily.
We write at roughly a 2nd–4th grade reading level, in short sentences, with natural language. No clickbait, no scare tactics, no marketing. That span is deliberately wide, because a 6-year-old is still decoding words while a 9-year-old reads fluently — so every story can also be read aloud, with each word highlighted as it’s spoken.
We’re honest that one reading level cannot serve that whole range well. Offering a simpler and a fuller version of each story is the most useful thing on our roadmap.
Hard limits — no exceptions for framing
Defence in depth — three independent layers
1. The rewrite stage
Before a story is written for kids, our AI is given the full safety rulebook in its instructions. If the source article is centred on something we don’t cover (an election, a war casualty, a gambling story), the rewrite is refused and the story never enters the pipeline.
2. A keyword backstop
The rewritten story is then scanned for unambiguous unsafe phrases — things like “head to the polls”, “polling station”, “the White House”, “casualty count”, “ballot box”. If any are present, the story is rejected even if our judgment AI missed it.
3. A separate kid-safety judge
A second AI then asks two questions in order. First, the affirmative gate: does this story’s subject belong to the kid world at all — nature, space, discovery, sports, arts, helpers, young achievers? If not, it fails, no matter how gently it’s written. Second, the harm screen: emotional safety, violence, mature content, gambling, politics, religious disputes, age-appropriateness, and the values the story would model for a child. A story is published only when both gates explicitly pass. If the judge can’t reach a confident decision, the story is held back — we never default to publishing.
Every decision is logged
Every safety decision — pass, block on content, or hold due to an infrastructure issue — is written to a moderation log with the headline, the dimension(s) that triggered the block, and the classifier’s reason in plain English. This lets us answer the question “why did this story run or not run?” for any article we’ve ever shown.
Regulators (notably UK Ofcom under the Online Safety Act) increasingly expect services used by children to demonstrate their moderation system is effective, not just that bad content gets removed later. The moderation log is how we do that.
So our rules don't quietly drift
We keep a fixed corpus of example stories that must pass (a library cat, a sea-turtle rescue, a young athlete’s comeback) and stories that must be blocked (election coverage, gambling framings, war casualty counts, civic-education framings of partisan politics).
Every time we change a rule — tighten one, loosen one, add a new category — we run the whole corpus through the safety system and confirm nothing slipped. If a story that should pass starts being blocked, or vice versa, we fix the rule before it ships.
What we collect, what we don't
The Scoop is designed for under-13 readers, so the site itself is built to know as little as possible:
The amended COPPA Rule took effect in June 2025, with full compliance required from April 2026. Our privacy notice sets out what is stored, where, and for how long.
What the rules ask for, and the choice we made
Children’s products are governed by a growing set of rules about privacy, safety and design. Most of them are about design choices rather than paperwork, so rather than list statutes, here is what each one asks for and what we actually did about it.
Children’s privacy — COPPA (US)
The rule governs collecting personal information from children under 13, so our answer is to not collect any. Kids never sign in. In The Scoop app, a child’s first name, avatar, reading progress, saved stories, poll answers and badges are stored on their own device and are never uploaded — a parent can see the full list and erase it from the Parent Zone. Accounts, email and sharing are parent-only. Our servers necessarily see an IP address when a page is requested; we use that to serve the page and nothing else — no profiles, no behavioural targeting, no sale or sharing of data.
Design that doesn’t pressure children — EU Digital Services Act Article 28, the UK ICO Children’s Code, and US state age-appropriate design codes
The European Commission’s July 2025 guidelines recommend that engagement mechanics such as streaks be turned off by default for minors, and several US states now prohibit “dark patterns” aimed at children. This one changed our product. We removed the daily reading streak: a streak’s pull comes from the fear of breaking it, and a seven-year-old doesn’t control their own schedule. What we show instead is the total number of stories finished — a number that only ever goes up, so there is nothing to lose and nothing to defend. A weekly view of days read is available, but a parent has to switch it on. There are no leaderboards, no comparison with other children, no autoplay, and no notifications that imply a child has fallen behind.
Apple’s Kids Category requirements
Anything meant for a grown-up sits behind a parental gate: a maths challenge that changes every time, then a four-digit code, and the zone re-locks the moment the app goes to the background. Every outbound link, the sign-in flow and the sharing tools live behind it. There is no advertising, nothing to buy, and no third-party analytics or advertising code anywhere in the app — including on the parent side, where the rules still apply.
Honest claims about AI — US FTC
The FTC has been active against companies making AI claims they can’t back up. Our stories carry the label “Checked for ages 6–9” rather than “AI checked & safe.” The checks described above are real and we stand behind them, but they are automated, and an unqualified promise of safety is not something any automated system can honestly make.
Usability research for children — Nielsen Norman Group
Not a regulation, but it settles arguments. Children react badly to content pitched even one grade off, so the app targets a narrow band and says so. Beginning readers can’t interpret elapsed time, so stories say “New today” rather than “18h ago.” They can’t spell reliably, so search forgives typos and browsing by picture comes first. They read icons literally, so buttons carry words. And because many of our readers are still learning to decode text, every story can be read aloud with the words highlighted as they are spoken.
What we don’t claim
This page describes our own practice. It isn’t a certification, no regulator has audited us, and none of it is legal advice. We’re a small team and the rules are moving quickly — if you think we’ve got something wrong, or that a claim here doesn’t match what the product actually does, please tell us and we’ll correct it. We update this page when our practice changes, not just when the law does.
Found a story that shouldn’t have made it through? Every story has a “Something not right? Tell us” link at the bottom — reports are anonymous, they go straight to our team, and we fix the rule, not just remove the story.