Research prototype. Synthetic data and simulated payments only — not an official or production security service. Limitations

Not sure it's real?
Check with Tumaini.

Voices can be cloned. Videos can be manipulated. Messages can pretend to come from people you trust.

Tumaini helps you check what you receive, understand the evidence and know what to do next.

See how Tumaini worksTry the live demo

Research prototype · synthetic data and simulated payments · not affiliated with any government entity or bank. The mobile companion is in development; the web demo works today.

Concept screens · Tumaini Assist, in development

It's getting harder to know what's real.

  • A voice message that sounds like someone you love.

    A few seconds of someone's voice is enough to copy it.

  • A video that appears to show an official announcement.

    Faces and voices can be changed to say things never said.

  • A message asking you to send money or share a code.

    It looks like it's from your bank, your boss or the government.

Three simple steps.

  1. 1

    Share what you're unsure about.

    Choose Tumaini when sharing supported content, or select something to check.

  2. 2

    Let Tumaini examine it.

    Tumaini checks available authenticity evidence and looks for suspicious signs.

  3. 3

    Know what to do next.

    Get a clear explanation and practical guidance.

More about how it works →

See how Tumaini helps in real life.

Pick a situation. Each one opens a real demonstration that runs on the live system with fictional people and organisations.

For anyone with family who might call for help

Family voice impersonation

Voice note · chat app · from “Omar” on a new number

“It's me, I lost my phone. Please send AED 5,000 today. Don't tell Dad.”

Fictional example

What Tumaini checks

  • Voice deepfake checkOpen AI model scores the recording — it can be wrong either wayReal check
  • What it asks you to doMoney, urgency, secrecy and a “new number” are foundReal check
  • Payment holdThe payment cannot go ahead until it is verifiedReal check
  • Confirm with the real OmarSent to the number you saved earlier — delivery is simulated on screenSimulated

Please check before sending money.

We cannot confirm that the speaker is your family member.

Next step: Call him using a number you already know.

Check your own voice note

Live runs use synthetic content in your own private, resettable workspace. No real money or organisations are involved.

Open the demonstration centre — all six situations →

Tumaini Assist · for people

A little help, right when you need it.

The companion app is designed to work through sharing. When something feels wrong, share it to Tumaini from the app it arrived in — Tumaini only sees what you choose to share.

Mobile companion in development

  1. 1Tap Share, choose Tumaini

  2. 2Say who it claims to be

  3. 3Tumaini checks it

  4. 4Know what to do

All four are concept screens of the app in development — not screenshots of a released app.

Tumaini for people

Tumaini Trust · for organisations

Help people recognise genuine communications.

Banks, government entities and businesses sign what they send. When someone checks a message, Tumaini can show it really came from them — and when it didn't.

Prove
Issue verifiable communications and manage your signing identity.
Protect
Add verification to payments and other sensitive steps.
Respond
See impersonation reports and failed checks as they happen.
For organisationsOpen the portal demo
Illustrative dashboard with a fictional bank. No real institution uses Tumaini.

Tumaini Core · for developers

Verification that fits into your application.

The same checks behind Tumaini Assist and Tumaini Trust are available through an API, so banks and apps can verify a message at the moment it matters.

Developer documentationSee a bank integration
Verify a message before payingLive public endpoint
POST /api/v1/verify/receipt
{ "receipt": "eyJhbGciOiJFZERTQSIs…",
  "intended_action": { "type": "payment",
    "payment": { "amount_minor": 15000, "currency": "AED", … } } }

// 200 OK
{ "outcome": "valid", "valid": true,
  "issuer": { "display_name": "Falcon Demo Bank", "status": "active" },
  "checks": [ { "id": "signature", "status": "pass" },
              { "id": "content_integrity", "status": "pass" },
              { "id": "scope", "status": "pass" }, … ] }

Clear answers. Honest limitations.

Tumaini keeps four kinds of evidence separate and tells you which is which.

  • 01

    AI analysis

    Does this look computer-made or suspicious?

    It estimates. It can be wrong in both directions.

  • 02

    Provenance

    Does the file carry a record of who made or edited it?

    Most genuine files have none. Missing ≠ fake.

  • 03

    Institutional verification

    Did this registered organisation sign exactly this message?

    It proves who sent it, not that the request is wise.

  • 04

    Independent confirmation

    Does the real person confirm it on a number you already had?

    It depends on contacts you saved beforehand.

No single AI result proves that something is real or fake. That is why payments wait for independent confirmation, whatever the score.

How it works technicallyLimitations

When in doubt, check with Tumaini.

Explore the demo For organisations