The problem

Seeing is no longer believing.

Generative media can fake a face, a voice, or a whole scene in seconds. The tools we have prove where a file came from, not whether the moment was real. That gap is where fraud lives.

Anyone can fake a face

Deepfakes are cheap, fast, and good enough to fool a screen. A convincing video is no longer evidence of anything.

A "verified" file can still be fake

Tools that tag a photo or video can show the file was not edited after the fact. They cannot show that what is in it really happened. Point a camera at a deepfake playing on a screen and you still get a clean, signed file of something fake.

Trust does not scale

Manual review is slow and subjective. Verification has to be something a machine can check and anyone can audit.

How it works

Five layers of truth.

One real event, witnessed five independent ways. A faker can spoof a single signal, never all five agreeing at the same instant, under a challenge nobody could predict. Two layers are hard gates; the more that agree, the higher the level. The proof carries fingerprints only, never your media.

The event

One live capture Recorded in-app, in the moment, with a random challenge fired mid-capture that no pre-recording can answer. live · on device

Five independent checks

Motion ↔ videoFootage moves exactly with the phone’s motion sensors.

defeats stapled deepfakes

Challenge responseAnswers a random live cue at the right millisecond.

● gatedefeats replays & pre-records

Depth livenessThe scene is 3D (LiDAR depth), not a flat screen.

● gatedefeats screen / green-screen

Pulse coherenceFace pulse matches the paired heart sensor, live.

defeats impersonation

Temporal consistencyClocks line up; nothing spliced or back-dated.

defeats doctored logs

The levels

L1

Livecaptured live, software-checked

L2

Verified devicegenuine app, real device

L3

Verified scenereal 3D depth, attested sensors

L4

Fully attestedfull attestation, pro grade

The more checks that attest, the higher the level. Both gates must hold to verify; if either fails, the proof is flagged. Web captures reach L1; higher levels need the native app and hardware attestation.

Fakes can copy one signal. Not five, all agreeing at the same instant, under a challenge nobody could predict.

A network, not a vault

Anchored on a public chain, checkable from anywhere.

Each proof is anchored on a public chain, written to validators around the world, so it can't be quietly changed and vIRL can't forge or alter it. Anyone can recompute the hash and check it against the public anchor. No account needed.

N. VirginiaLos AngelesFrankfurtLondonTokyoSingaporeSão PauloSydney

0ways to alter a proof

<1sto check a proof

Publicon-chain anchor

The output

Proof you can see.

When a capture verifies, vIRL stamps a clean credential onto the photo or video and anchors its fingerprint. The mark travels with the file, and anyone who sees it can open the full proof to check that it was captured live, by a real person, and has not changed since.

✓A clean keepsake frame with a scannable QR, not a watermark slapped across your photo.
✓Carries the verification level and date, backed by the on-chain proof.
✓Saves straight to your photo library, ready to share.

A real vIRL keepsake: captured live, stamped with the vIRL mark and a scannable QR that opens the on-chain proof.

✓ Real capture, verified on devnet. Scan the QR to open the proof.

Honest status

What works today, and what is next.

This is an early prototype. We would rather show you exactly where the line is than pretend the whole vision already ships.

Working in this prototype Live

✓Live liveness capture using phone motion and the camera.
✓Random challenge that blocks pre-recorded and replayed clips.
✓Real SHA-256 proof manifests generated on the device.
✓On-chain anchoring of the proof hash to Solana devnet, with a shareable transaction link. vIRL sponsors the fee, so no wallet is needed.
✓Open verifier page that reads a real anchored proof back off-chain.
✓Open-source verification core, Apache-2.0: the proof format and a runnable verifier are public, so anyone can check our work. View the code →
✓English, Spanish, and Italian for testers.

On the roadmap Building

○Mainnet anchoring and Merkle batching, after a security review.
○Hardware sensor signing and a trusted enclave.
○LiDAR depth to defeat screen and green-screen replays.
○Heart-rate liveness from video, matched to a wearable.
○A native iOS capture app and a confidence score.

Where it matters

Proof for moments that matter.

vIRL proves the capture was live, a real person in the moment. What you do with that proof is up to you.

Insurance & claims

Document damage the moment it happens. A live, time-stamped record you can hand an adjuster, not stock or recycled footage.

Buying & selling

Show the item is in your hands right now, captured live, not a saved or downloaded photo.

Journalism & eyewitness

Prove footage was shot live, by a real person, not recycled or a pre-recorded clip.

Real person, real profile

Prove there’s a real, live human behind the account, not a stolen photo or a pre-recorded clip.

See it in action

Capture, prove, share.

No wallet, no setup. Watch a real capture, from live moment to shareable proof.

Turn your sound on for the walkthrough, or use the mute button in the player.

Verified in the wild

Real moments, checkable by anyone.

Captures people made and shared. Every card links to its live on-chain proof, so you don't have to take our word for it. Open any one and check it yourself.

Try it now

Prove your next moment.

Capture a live moment in your browser, or open the verifier to see how a proof is checked.

Live capture

Patent pending. US provisional 63/948,969. Raw media and biometrics never touch the chain.

Proof you were really there. Proof a real person filmed this. Proof it was captured live. Proof it’s in your hands right now. Proof you filmed it yourself.

Anchored & tamper-evident