CCPACCPA privacy rightsAccess and deletion requests honoured.How we handle data
NVIDIA InceptionMember of NVIDIA’s program for startups.
01 · The problem
Online, you can be anyone.
Passwords and one-time codes prove someone has the right device. They don’t prove who is holding it. Now AI makes convincing fake IDs and faces easy to produce, and when the request is to move money or take over an account, that gap is where fraud gets in.
A borrowed identity
Someone uses another person’s details and photo to pass as them.
A deepfake ID
An AI-generated ID image that looks real on a screen, made to pass a quick look.
A face that isn’t theirs
A mask, a printed photo, or a replayed video held up to the camera.
What independent testing found
Today’s identity checks miss too much.
The US Department of Homeland Security tested commercial identity verification systems against fake IDs, spoofed faces and real people. The typical system:
1 in 8
fake IDs got through
87% of 1,900+ fake IDs from all 50 states were caught.
How Stile answers it
A document-fraud model, custom-trained with NVIDIA AI tools on thousands of identity documents, catches the subtle edits and AI-generated alterations that template-based checks can miss. Barcodes are checked against the printed details.
1 in 2
of the strongest face spoofs got through
53% of the most successful presentation attack was caught.
How Stile answers it
Liveness detection, lab-tested by iBeta to ISO/IEC 30107-3 Level 1 and 2, confirms a live person is present, not a photo, a screen or a mask.
1 in 4
genuine people were turned away
75% of genuine users passed every step.
How Stile answers it
Image capture and text reading are tuned for everyday phone photos, using Microsoft Azure for reading documents and AWS for face matching. A blurry or glared shot is retaken in the same session, not rejected.
Finding 1 of 3
Next, DHS is adding a test for AI deepfakes of selfies and ID photos (2027).
Source: DHS Science and Technology Directorate, Remote Identity Validation Rally (RIVR-1), median results for tested commercial systems, presented September 30, 2026.
02 · How Stile protects you
Catch the fake. Keep no copy.
1234
Fake-document detection
Four checks run when someone verifies with a photo ID.
1AuthenticityIs the card genuine?
2Barcode cross-checkDo the codes match the print?
3Face matchIs the selfie the person on the ID?
4LivenessA live person, not a photo, screen or mask
An identity verification platform checks that a person is who they claim to be, usually by reading a government ID and matching it to a live selfie. Stile does this online or in person, at the moment of a risky request such as a payment-detail change, an account recovery or a high-value pickup, and sends your system a signed result.
The person opens a link or scans a QR code, photographs their ID or shares a mobile driver’s license, and takes a selfie. Stile checks the document, matches the face and confirms a live person, then sends the result to your team. ID images are deleted once there’s a final decision.
$0.25 per full check, covering the ID document, biometric face match, age and liveness. The first 1,000 completed checks are free every month, and there is no monthly base, per-seat fee or minimum commitment. Volume rates are on the pricing page.
Yes. Stile publishes its price: $0.25 for a full check, with 1,000 free checks a month. For comparison, Stripe Identity's published list price is $1.50 per completed verification. Stile's liveness technology is lab-tested to ISO PAD Level 1 and 2, and ID images are deleted once a verification reaches its final decision.
Yes. A sandbox organization runs up to 500 test checks a month for free, and the first 1,000 production checks are free every month on a live organization, with no card on file.
No. There's no monthly base, no per-seat fee, and no minimum commitment. A month with no checks costs nothing.
A phone with a camera and a physical ID such as a driver license. On a computer, you scan a QR code to continue on your phone.
The trial is a real verification, and Stile’s default retention applies: ID images and the details read from them are deleted once it reaches a final decision. If it can’t be decided automatically, a member of Stile’s staff may review the ID before it’s decided.
Send a hosted link with no code, show a QR code at the counter, or add the Verify with Stile button to your app. Your system receives a signed result, and the developer docs cover each path.