Verified
The registered person approved this exact request on their trusted device. You can act on it.
Casi d’uso
Deepfakes
A convincing imitation of a voice or a face still does not hold the credential. Verify the person, not the recording.
The problem
Recognising a fake is becoming a losing game
Synthetic voice needs seconds of reference audio. Video calls can be driven live. "Does this sound like them?" is a test that gets easier to pass every year, and the people most often targeted are the least equipped to run it.
What changes
Ask something a model cannot answer
Move the question from what the media looks like to what the person can prove. A perfect imitation of a voice does not carry the private key held inside the real person's phone, and cannot produce a signature over the exact request you are asking about.
How this works
Verifying the person behind the call
- A call or clip makes a request It sounds like someone you know, or someone with authority over you.
- You ask REAL AUTH instead of the caller The verification is sent to the account the claimed person actually registered, not to the number or address that contacted you.
- Their real device asks them They see the request in full and answer on hardware only they hold.
- You see the answer, or the absence of one Verified, rejected, not verified, or no response. Only the first one means go ahead.
Possible outcomes
Four answers, and one of them is silence
A verification request has more than two endings. Treating anything other than an explicit approval as approval is how verification systems fail people.
Rejected
The registered person saw this request and refused it. Treat it as fraudulent and report it.
Not verified
Nothing was proven, either way. Do not act on the request until you confirm it independently.
No response
Nobody answered. This is not approval. Silence is the outcome an attacker hopes you will read as consent.
No response and Not verified are never shown as approval, and they never turn into approval by waiting.
Honest limits
What this does not fix
REAL AUTH makes no attempt to detect whether media is synthetic, and no claim to be able to. It also cannot help if the claimed person has no REAL AUTH account, or if their device itself is in someone else's hands. It changes the question being asked; it does not win the arms race over generated media.
Related use cases
Family scams
"Dad, I had an accident and I need money now." Ask your daughter, instead of trusting a voice that sounds like…
Scopri di più →Business impersonation
CEO fraud, fake suppliers, fake support desks. Check that the person is real and that they hold the authority…
Scopri di più →Phishing and impersonation
Verify the request through a channel the attacker does not control, instead of judging whether a message looks…
Scopri di più →Try it against your own application
A free sandbox, real flows, and no card required.