Check it before you trust or share it.

Health misinformation reaches young people in Benin before they go looking for it. Loupe answers it inside the apps they already use — WhatsApp, Messenger, Instagram — and in public, under the post itself.

Try it on Telegram The data behind it

Pilot in Benin · In French, with Fon and Yoruba audio

The problem

0%

of the young people we surveyed meet health claims while scrolling — without ever searching for them.

She is twenty, in Cotonou, scrolling between two classes. A woman in a white coat says, with total confidence, that contraception leaves young women sterile. That evening a friend forwards her the same video. She stops her contraception, and asks no one.

Nothing in that started with a question. The claim found her.

Why now

Generative AI made a convincing fake doctor cheap.

0%

of citizens worldwide are worried about the impact of online disinformationUNESCO / IPSOS, 2023

0%

of Beninese get news regularly through social media — eight times more than in 2014Afrobarometer, 2026

0

young Beninese were trained as fact-checkers by WHO and UNFPA in 2022, reaching 2,000 peersWHO Africa, 2022

Benin has already used youth and social media against health misinformation — during one epidemic. Loupe extends that approach to the everyday: contraception, infections, miracle cures, self-medication.

How it works

Wherever the claim reached them.

Step one

Forward it

A message, a screenshot, a voice note or a video goes to Loupe — on WhatsApp, or in the Messenger, Instagram or Facebook inbox they were already in. Nothing to install, no website to visit, no account to create.

Step two

Read the verdict

The answer comes back in your own private chat with Loupe, in plain language. No percentages, no scores. Loupe never enters the conversation the content came from, and the friend who sent it is never notified.

Step three

See the source

When our verified library of WHO and Benin health-authority sources covers the claim, the verdict cites it. When it does not, the verdict says so plainly and sends you to a health professional.

Or, in public

Tag Loupe under the post

Forwarding is private, and nobody is told. But a young person can instead tag Loupe in a comment under a Facebook or Instagram post, and Loupe answers in that comment thread — so everyone who saw the claim reads the verdict too. One rumour, answered once, for everybody.

Try the working prototype on Telegram →

Loupe is a verifier.
Not a health adviser.

It answers no health questions. It gives no diagnosis and no treatment advice. It takes a claim that has already reached a young person and says whether it is reliable — and why.

We asked first

We surveyed 46 young people in Benin.

Where young people meet health information

Share of 46 respondents naming each platform. Multiple answers allowed.

WhatsApp93% Facebook93% TikTok89% YouTube11% Search engines2% 0 50% 100%

One respondent in forty-six found health information through a search engine. Forty-three found it on WhatsApp.

0%

would use a free tool that tells them whether a health claim is reliable, and why

0%

have already shared health information without checking it

0%

say a simple explanation is what would help most — ahead of everything else

Asked what would help most, the first answer was not more information. It was a simple explanation. So every verdict is written in plain language.

How a verdict is built

No verdict reaches a user on one model's word.

01

Retrieval

Searches a curated library of WHO and Benin health-authority documents, selected and validated by our Health Content Lead, and collects the passages relevant to the claim.

02

Writing

Drafts the verdict in plain language from those passages, and cites them. When the library holds nothing on the claim, it says so explicitly rather than guessing.

03

Checking

Verifies that every statement is supported by a cited passage, that every source exists, and that nothing drifted into diagnosis or treatment advice. A draft that fails is rewritten once.

Verification is the entry point

The goal is the skill.

Real or Fake?

A daily game comparing authentic and manipulated health content. It trains the eye on fake authority, fabricated testimonials and synthetic media.

Trust Circle

One tap brings a friend, or a trained peer educator, into the conversation. Human judgment alongside the AI. You always decide.

Peer workshops

Trained youth ambassadors run sessions in universities, youth centres and community spaces, and recruit the next ambassadors.

Public trend alerts

Anonymised claims let us see the same rumour arriving again and again. We publish a short public warning for each one, so a viral falsehood becomes a collective lesson.

The day a young person checks the next claim without us, we have done our job.

Safety and privacy

A health verdict carries a duty.

  • No diagnosis, ever

    Loupe assesses whether a claim is reliable. Whenever a claim touches a symptom, a medication or an emergency, it advises consulting a health professional.

  • Reviewed by health professionals

    Content is defined by our Health Content Lead, trained in general medicine, and reviewed weekly by three confirmed health professionals from the Comè health zone.

  • Nothing kept that we do not need

    Messages, images, voice notes and phone numbers are used to produce the verdict, then discarded. WhatsApp users need no account.

  • Claims anonymised before storage

    A checked claim is reduced to a short neutral statement, stripped of names and identifying detail, and linked to no user. That is what lets us spot recurring rumours.

  • The AI states its limits

    Loupe says clearly when it is uncertain, and never presents a verdict as medical certainty. Any user can flag a verdict in one tap.

  • Peer educators are accountable

    Trained peer educators sign a code of conduct covering confidentiality, the boundary against medical advice, and a named referral point.

Our data handling is advised by

  • Dr Geraud PAZOUN — Cybersecurity
  • Sèdofi Hope LANMAYIKPOHOUE — Security Engineer

They advise on data minimisation, on encryption in transit and at rest, and on how the messages, images and voice notes people send us are handled — the areas where a service like ours can do the most harm if it is built carelessly.

Six-month pilot

What we are holding ourselves to.

0

young people reached

0

health claims verified

0

trained peer educators

0

peer workshops delivered

Progress is measured by a short diagnostic taken before and after each cycle — and by how far young people get in the game without our help, which is the only honest test of whether they need us less.

Who builds it

Five young people from Benin. Three public institutions.

  • Josaphat Adjele

    Team Lead & Full-Stack Developer

    ENSGMM, UNSTIM

  • Hortis Houenou

    AI & Data Engineer

    ENSGMM, UNSTIM

  • Abel Kpokouta

    Community Manager

    ENSGMM, UNSTIM

  • Mardochée Adjele

    Statistician

    ENSPD, Université de Parakou

  • Florida Déo Gratias NASSOUHOUEDE

    Health Content Lead

    Faculté de Médecine, Université de Parakou

Our health content is reviewed by

  • Dr Amour S. FAGNIHOUN — Coordinating Physician, Comè health zone
  • Juliette D. DJISSONON — Head of Obstetric Care, Comè health zone
  • Rock SAVI — Epidemiologist