Check the person and the papers at onboarding.
One call to /v1/verify checks a selfie for deepfakes, matches it to the ID photo, validates Aadhaar, PAN and passport numbers by their own arithmetic, and re-adds a bank statement's balances. You get a pass, review or flag decision with the evidence behind it.
Uses the same credits as image authenticity checks: one credit per identity check, with 100 free on sign-up.
Send what you have. Each part you include adds its own check to the same decision.
What one identity check covers
| Check | What it does | What it catches | Detail |
|---|---|---|---|
| Deepfake selfie | Scores the selfie for face swapping and full synthetic generation. | A victim's face mapped onto an applicant, or a person who does not exist. | Deepfake detection |
| Face match | Compares the selfie with the photo on the submitted ID. | Borrowed and stolen documents. | Identity verification |
| Aadhaar | Checks the number against its Verhoeff checksum. | Made-up and mistyped Aadhaar numbers. | Aadhaar and PAN |
| PAN | Checks the number against the PAN format. | Fabricated and malformed PAN numbers. | Aadhaar and PAN |
| Passport | Reads the machine-readable zone and recomputes its ICAO 9303 check digits. | Edited and forged passports. | Document fraud |
| Bank statement | Re-adds the running balance line by line, and reconciles salary slip earnings, deductions and net pay. | Edited balances and inflated income. | Bank statements |
| Liveness | A passive presentation-attack signal from the selfie. | A photo of a photo or a screen held up to the camera. | Video KYC |
| Injection | Reads device signals from our capture SDK: virtual cameras, emulators and screen feeds. | A recorded or synthetic video fed into the camera pipeline. | Video KYC |
Arithmetic can accuse. A model alone cannot.
How a decision is reached
A genuine customer is never rejected on a model's guess. The rules that turn the signals into an outcome are fixed and the same for every customer.
Every check is clean
The selfie reads as a real face, it matches the ID, the document numbers satisfy their checks and the statement adds up.
A person decides
A signal is uncertain, only one family of signals raises a concern, a submitted video could not be analysed, or the capture SDK reports an injection signal. None of these can pass on their own.
Proof, or agreement
A checksum fails or a statement does not add up, which is proof. Or at least two independent families of model signals agree that the selfie is fake.
One request from your backend
Post the selfie, the ID image or a bank statement PDF, in any combination. Add the capture SDK's device signals when the selfie was taken in your web or mobile flow.
| Response field | What it holds |
|---|---|
decision | pass, review or flag. |
risk_score | 0 to 100. Higher means more likely fraud. |
selfie, liveness | The deepfake and presentation-attack readings for the selfie. |
face_match | Whether the selfie matches the ID photo, with its probability. |
document | The checksum, format or reconciliation result, and whether it was decided by arithmetic. |
injection | The capture SDK's verdict and its reasons. |
evidence | Each finding behind the decision, for your reviewer. |
recommendation | What to do next, in one sentence. |
Full request and response details are in the API reference. The browser capture SDK runs on your onboarding page; see the capture SDK sample.
curl https://truesecurix.com/v1/verify \ -H "X-API-Key: $TRUESECURIX_API_KEY" \ -H "Idempotency-Key: app-90317-kyc" \ -F "selfie=@selfie.jpg" \ -F "document=@aadhaar.jpg" \ -F "reference=app-90317" \ -F "device_signals=$SIGNALS_JSON" # A bank statement goes in the same field curl https://truesecurix.com/v1/verify \ -H "X-API-Key: $TRUESECURIX_API_KEY" \ -F "document=@statement.pdf"
What a compliance review asks, answered before it is asked.
What we do and do not claim
No government database lookup
We validate a document's own checksum and format and match its photo to the selfie. We do not claim a UIDAI or Income Tax database lookup, or any government authorisation we do not hold.
A layer beside your video KYC
The deepfake and injection checks run alongside your existing video KYC or liveness provider. We do not claim RBI empanelment.
Device signals advise
Capture SDK signals raise the bar against casual injection. A determined attacker can spoof any client value, so on their own these signals can send a capture to review but never reject it.
Files are not kept
Selfies, ID images and statements are analysed in memory and never stored. Aadhaar numbers are masked in the evidence. We keep a SHA-256 fingerprint and the result.
Each check in detail
Deepfake detection API
Face-swapped and fully synthetic selfies at onboarding.
Deepfake-resistant video KYC
Injected and AI-generated faces in a V-CIP flow.
Aadhaar and PAN verification
Verhoeff checksum, PAN format and selfie to ID face match.
Document fraud detection
Aadhaar, PAN and passport MRZ checked by their own arithmetic.
Bank statement fraud detection
Running balances and salary slips reconciled line by line.
Identity verification API for India
All of the above in a single onboarding decision.
Test it on your own onboarding files
Create an account for 100 free checks, no card needed. Test keys return simulated results at no charge while you build the integration.