Security and privacy
Prove it without giving it away.
Helix checks what a service needs to know, right on your device. Your voice recording never leaves your phone or computer.
The one promise
All audio processing happens on your device. Your voice recording is never uploaded or shared.
- 01
You say four numbers.
It takes about four seconds.
- 02
Your device checks it.
Is this a real, live human voice? And what does the service need to know?
- 03
Your recording stays put.
The service gets its answer. The recording never leaves the device it was made on.
Prove only what's needed
The service gets the answer. Not your voice.
In plain terms
What leaves your device, and what doesn't.
| Stays on your device | Can be sent to Helix |
|---|---|
| Stays on your deviceYour voice recording | Can be sent to HelixThe result, for example 18+ |
| Stays on your deviceRaw audio, always | Can be sent to HelixA voice signature (a set of numbers), only for features that recognize you over time, like voice MFA (signing in with your voice) |
What's a voice signature? A set of numbers that lets a service recognize you next time. It isn't a recording.
When a service uses passkeys, your device also sends a signal confirming it's a real device. It doesn't include your voice.
For business
Collect less. Know more.
Get the answer your trust and safety team needs without taking on voice data you'd have to protect.
Lower your liability
No audio reaches your servers or ours, so there's nothing to store or breach.
Turn on only what you need
Age checks, voice MFA and synthetic speech detection are separate settings.
Synthetic speech and replay detection
Deepfake and synthetic voice checks are built into every checkpoint.
Compliance
Independently certified by the Age Check Certification Scheme (ACCS) for ISO/IEC 27566-1, GDPR, and the New York SAFE for Kids Act.
Using a service with Helix?
See what's checked, what's shared, and what your rights are.
Building a platform?
See how Helix fits your trust and safety stack.