A 12-Year-Old Drew a Mustache With Eyeliner and Fooled the UK's Age Verification AI

:disguised_face: A 12-Year-Old Drew a Mustache With Eyeliner and Fooled the UK’s Age Verification AI

Governments spent millions building “facial age estimation” — kids cracked it with a $3 eyebrow pencil

46% of UK kids say age checks are “easy to bypass.” 32% have already done it. And 17% of parents are literally helping them cheat.

A nonprofit called Internet Matters surveyed over 1,000 UK children and parents about the shiny new age verification systems that the UK’s Online Safety Act is forcing onto websites. The results are… honestly, exactly what anyone who’s ever met a teenager would expect.

Mustache Disguise


🧩 Dumb Mode Dictionary
Term What It Actually Means
Facial age estimation Software that looks at your face through a webcam and guesses how old you are. Like a bouncer, but worse at its job.
Online Safety Act (UK) A British law that says websites must verify users are old enough before showing them adult content or letting them sign up.
Age gating Putting a “are you 18?” wall between a user and content. Usually a birthday dropdown menu that literally nobody fills in honestly.
Yoti One of the main companies making the face-scanning age check tech. They sell it to platforms as a plug-and-play solution.
Biometric verification Using your body (face, fingerprint, eye scan) to prove who you are. Sounds sci-fi. Gets beaten by cosmetics.
📰 What Actually Happened

One mother told researchers she caught her 12-year-old son using an eyebrow pencil to draw a mustache on his face before sitting down at the webcam.

The facial age estimation software looked at this child’s drawn-on facial hair and said: “Yep, you’re 15.” Verified. Done.

He wasn’t an edge case. The survey by Internet Matters found that nearly a third of kids admitted to bypassing age checks — and those are just the ones who told the truth on the survey.

🎭 The Full Cheat Sheet Kids Are Using

Beyond the legendary mustache move, kids reported a whole menu of bypasses:

  • Drawing facial hair with makeup pencils (the classic)
  • Pointing their webcam at a video game character — some face-scan systems apparently can’t tell the difference between a real person and a rendered NPC
  • Pulling weird or exaggerated faces — distorting their features enough that the AI’s age estimate jumps up
  • Typing in a fake birthday — the oldest trick in the book, still works on most sites
  • Using a parent’s or sibling’s ID card when a photo ID upload is required

Honestly, the video game character one is sending me. A kid pointed a webcam at Master Chief or whatever and the AI went “ah yes, a responsible adult.”

📊 The Receipts
Stat Number
Kids who say age checks are easy to beat 46%
Kids who’ve actually bypassed one 32%
Kids who say it’s difficult to fool 17%
Parents who actively helped kids cheat 17%
Parents who saw their kid cheating and looked the other way 9%
Kids who encountered harmful content recently 49%

That parents stat is wild. Nearly 1 in 5 parents are sitting there going “here, use my driver’s license, I want to watch the match in peace.”

🗣️ What People Are Saying

Rachel Huggins, CEO of Internet Matters:

“Stronger action is needed from both government and industry to ensure that children can only access online services appropriate for their age.”

Security researchers have been saying for years that facial age estimation is fundamentally flawed — the technology guesses based on visual cues like jaw shape, wrinkles, and yes, facial hair. So any kid who can make themselves look slightly older breaks the system.

The EFF (Electronic Frontier Foundation) has argued these age gates create massive privacy risks — you’re handing your biometric face data to random companies just to prove you’re old enough — and they don’t even work.

🔍 Why This Keeps Failing

The core problem is simple: there are only a few ways to verify age online, and they all suck.

  • Birthday dropdowns — everyone lies. Always have, always will.
  • Face scanning (age estimation) — guesses your age from your face. Gets fooled by makeup, lighting, camera angles, and apparently video game characters.
  • ID upload — forces people to hand over government ID to private companies. Massive privacy nightmare. And kids just use their parents’ IDs anyway.
  • Credit card checks — not every adult has a credit card, and prepaid cards have no age attached.

Governments keep trying to square this circle. Australia tried banning under-16s from social media — 60% of teens still had access within months. The UK’s Online Safety Act mandates platforms use “age assurance” but doesn’t specify how, which is how we ended up with AI that gets defeated by an eyebrow pencil.


Cool. Face-scan AI can’t tell a 12-year-old from a 15-year-old with $3 worth of makeup. Now What the Hell Do We Do? (⊙_⊙)

Use Case Mustache

🕳️ The Age Gate Auditor

Platforms using facial age estimation need to know exactly how bad their systems are before regulators find out. But they can’t exactly hire 12-year-olds to test it (legal issues). There’s a gap here: independent age-verification testing services that use synthetic faces (AI-generated at different ages, with and without modifications like drawn-on facial hair) to stress-test these systems.

You don’t need to build the face-scan tech. You need to break it, document how, and sell the report to the companies who deployed it before Ofcom (the UK regulator) fines them.

:brain: Example: A 24-year-old pen tester in Estonia generates 500 synthetic faces aged 10-17 using free tools like StyleGAN, adds drawn-on mustaches in Photoshop, runs them through 3 major age-estimation APIs, publishes a failure-rate report. Sells it to 2 UK ed-tech companies for £3,000 each as a “compliance pre-audit.”

:chart_increasing: Timeline: First report in 7 days. Steady income for 6-12 months while Online Safety Act enforcement ramps up. Dies when platforms switch to better tech (or give up entirely).

🪟 The Patch Window Panic Pack

Right now, platforms are scrambling to comply with the UK’s Online Safety Act, and Ofcom is starting enforcement actions. Every website that serves UK users and hosts age-restricted content needs age verification yesterday. Most of them have no idea how to implement it.

The play: create a “compliance starter pack” — a dead-simple guide + template that walks small-to-mid website owners through which age verification provider to pick, how to integrate their API, and how to document it for Ofcom. Sell it as a one-time purchase. You’re not selling the tech, you’re selling the panic.

:brain: Example: A 21-year-old web dev in the Philippines reads Ofcom’s age assurance guidance docs, builds a Notion template with step-by-step integration guides for the 3 biggest age verification APIs (Yoti, VerifyMyAge, AgeChecked). Sells it on Gumroad for £49. Moves 200 copies in the first month from panicked UK forum owners.

:chart_increasing: Timeline: First sale within 4 days. Peak window is the next 3-4 months while enforcement is fresh and site owners are scared. Drops off as bigger compliance firms catch up.

📡 The Parental Bypass Tracker

Here’s the weird angle nobody’s talking about: 17% of parents are helping their kids bypass age checks. That’s not a bug — it’s a market signal. Parents want to supervise their kids’ internet access themselves, not outsource it to a broken face scanner.

The play: a browser extension or simple web dashboard that lets parents set custom content rules per kid — basically giving parents the keys back instead of trusting a dumb AI. Position it as “age verification that actually respects families” and market it directly in UK parenting Facebook groups and Mumsnet.

:brain: Example: A 26-year-old developer in Romania builds a Chrome/Firefox extension using WebExtensions API that lets parents whitelist/blacklist sites per child profile, logs what sites kids visit (without sending data to any server — fully local). Charges £2.99/month. Picks up 1,500 subscribers in 2 months from Mumsnet recommendations alone.

:chart_increasing: Timeline: MVP (basic working version) in 10 days. First paying users in 2 weeks. Sustainable as long as age-verification debate continues — which is basically forever.

🎣 The Synthetic Face Firehose

Companies building age estimation AI need training data — specifically, they need thousands of faces at known ages. But child face data is (rightfully) extremely restricted. Here’s the loophole: synthetic child face datasets generated by AI, where no real child’s photo is used, exist in a legal gray area that’s way easier to sell.

You generate realistic-but-fake face images at specific ages using open-source models, label them, and sell the dataset to age-estimation companies who are desperate for training data but can’t legally scrape real kids’ photos.

:brain: Example: A 23-year-old ML student in Poland uses this FFHQ-Aging dataset methodology to generate 10,000 synthetic faces labeled by estimated age bracket (8-10, 11-13, 14-16, 17-19). Sells the dataset to 2 age-verification startups for $5,000 each. Entire generation process runs on a rented GPU for under $50.

:chart_increasing: Timeline: Dataset ready in 5 days. First sale in 2-3 weeks (cold email age-verification companies listed on Ofcom’s registry). This is a repeating sale — each new client pays full price for the same dataset.

🎰 The Age Gate Escape Room

Okay but seriously — this entire story is comedy gold. A kid beat government-mandated AI with an eyebrow pencil. That’s practically performance art. The play: turn it into actual content.

Build a YouTube/TikTok series called something like “Can It Pass?” where you (as an adult, legally) test different age verification systems with increasingly absurd disguises — fake mustaches, wigs, masks, face paint, cosplay. It’s entertaining AND educational (shows how bad these systems are). Media outlets are already writing about this, so the SEO is primed.

:brain: Example: A 19-year-old in the UK films herself testing 5 different age verification tools while wearing different silly disguises. First video gets 340K views because every parent and tech person shares it. Gets picked up by BBC Click for a segment. Brand deals from VPN companies follow within weeks.

:chart_increasing: Timeline: First video up in 2 days. Viral window is NOW — this story is trending. Content works for 2-3 months before the joke gets old. Pivot to broader “testing security systems” format to keep it going.

🛠️ Follow-Up Actions
Want To… Do This
Understand the UK law Read Ofcom’s age assurance guidance — it’s the actual rules platforms must follow
See which age verification tools exist Check Yoti, VerifyMyAge, and AgeChecked — the big three in the UK market
Test age estimation yourself Try Yoti’s free age estimation demo on their site — see how it guesses YOUR age
Read the full survey Look up Internet Matters’ 2026 Online Safety Act Impact report on internetmatters.org
Understand the privacy side Read EFF’s guide on navigating age assurance systems

:high_voltage: Quick Hits

Want… Do…
:magnifying_glass_tilted_left: See how age estimation AI works Try Yoti’s age scanner demo — test it with sunglasses, hats, makeup
:shield: Protect your own kids properly Use a local-only content filter (no data sent to companies) instead of trusting face scans
:money_bag: Cash in on the compliance panic Build age-verification integration guides for small UK site owners before Ofcom starts fining
:brain: Go deeper on the privacy angle Read EFF’s age assurance breakdown
:face_with_steam_from_nose: See how Australia’s ban went Check the Fortune report — 60% bypass rate in 5 months

Governments built a billion-dollar digital bouncer. A 12-year-old beat it with his mom’s eyebrow pencil. And honestly? That tells you everything you need to know about who’s really running the internet.

2 Likes