Persona assembled → Records checked and consistent → AI-generated face supplied → Liveness prompt passed → Account opened → Credit built over time
Each stage looks valid. The authenticity of the face was never tested.
Synthetic identity / NeuroVerify
Deepfake Detection API for Synthetic Identity Fraud
Synthetic identity fraud builds a customer from real and invented data and gives it a face that was never a living person. NeuroVerify checks the face in your onboarding media for deepfakes and manipulation beyond liveness and returns one unified risk score with evidence.
It runs behind the selfie or video step your onboarding already has: no new step for genuine customers, one API call.
NeuroVerify does not detect synthetic data, verify records or decide on an application; it checks the face media, and your policy decides.
Powered byNeuroVerifyExplore the product
- #1 of 36on the public VendorBench-100
- <200 msp95 on eligible image requests
- 6 → 1six checks, one unified score
- 0new steps for your customer
The gap
Every record can be valid while the person does not exist.
Synthetic identity fraud is not a stolen wallet. A ring assembles a persona from real and invented parts: a valid identifier fragment, a real address, a mule bank account, a new phone number. Then it needs a face to complete the application. Generative tools now supply that face on demand.
Why standard checks miss it
Registry and bureau checks look for stolen identities, not invented ones with enough real data. Face match works if the face matches itself. Liveness can be passed by an injected or replayed generated face. See synthetic identity fraud for the wider picture.
The pattern often plays out slowly. A synthetic customer opens small accounts, builds a history, then borrows or cashes out. The best time to spot it is at onboarding, from the face itself.
The same face can also appear under different names. A fraud ring may generate several synthetic faces and run them across institutions, which makes consistency of the media an important signal.
Threats
How synthetic faces enter onboarding.
These are techniques, not claims about any institution. Each gives an invented person a convincing face.
01
Fully generated face as the selfie
A generated portrait, often with a matching ID-style photo, is submitted as a live capture.
02
Face swap onto a real-person source
A real person's video or image is altered so the face matches a constructed persona.
03
Virtual camera injection of a prepared face
A prepared or live-rendered face is fed in as the camera stream, so the SDK sees a normal video.
04
Reuse of one face across personas
The same synthetic face is submitted with different names and records, each looking clean alone.
05
Mule account enrolment with synthetic media
Accounts are opened for moving funds, using generated faces that no one can trace to a real person.
06
Slow-build synthetic customer
A persona keeps small, clean balances for months, then borrows or cashes out when limits rise.
The earlier a synthetic face is flagged, the less history it has to hide behind.
Where NeuroVerify fits
Test the face before the identity builds history.
NeuroVerify sits behind the selfie or video capture in your onboarding. You send the media you already collect. It returns one unified risk score with evidence from six signals, including checks for AI-generated imagery and pixel-level manipulation, and your policy decides whether to approve, step up or decline.
NeuroVerify gives you evidence about the media, not about the persona's records. Pair it with your data checks, device signals and consortium data.
NeuroVerify Unified Trust LayerSix parallel analysis layers
- Visual Safety Screening
- Live Presence Verification
- Face Manipulation Detection
- AI Generative Model Analysis
- AI Pixel-Level Manipulation
- AI Content Compliance
Send the onboarding selfie or video
Pass the media from your existing capture step to the API with an SDK or REST.
Read the unified risk score
Get one score with signal-level evidence, including generative and pixel-level indicators.
Apply your risk policy
Approve, step up with another check, hold for review or decline, by product and exposure.
Feed evidence into fraud operations
Store the output and link repeated media patterns across applications.
What you get
What fraud and risk teams gain.
A check aimed at the synthetic face
Detect AI-generated and manipulated faces that records and liveness can miss.
Earlier detection
Flag the face at onboarding, before a synthetic customer builds history.
Evidence for investigations
Signal-level output helps analysts group related applications.
Fits existing KYC
Keep your document, liveness and bureau vendors; add one API beside them.
Public benchmark result
First of 36 systems on VendorBench-100, with MCC 0.876 versus 0.646 for the next-best vendor.
Questions
What teams ask before they integrate.
How does NeuroVerify fit into your onboarding and account-opening flow?
It sits behind the selfie or face-capture step you already run. You send the image, video or audio you already collect to one API call, and the response comes back with a unified risk score and signal-level evidence. The end user sees no new step.
Integration uses native mobile SDKs (Android/Kotlin, iOS/Swift, Flutter), backend SDKs (Python, TypeScript/Node.js, Go, Java) or the REST API. Two lines to integrate and live in under 24 hours are product claims; your own environment and review process set the real timeline.
We already run liveness. Why add deepfake detection?
Liveness asks whether a live person appears to be in front of the camera. It was designed around printed photos, screen replays and masks. It does not ask how the media was produced, so a face swap, a fully synthetic face or an injected stream can satisfy it.
NeuroVerify checks for deepfakes and manipulation beyond liveness. Keep your liveness vendor and add the authenticity evidence next to it. See deepfake detection vs liveness detection.
What does NeuroVerify return, and who makes the decision?
One unified risk score, plus evidence from six signals: Visual Safety Screening, Live Presence Verification, Face Manipulation Detection, AI Generative Model Analysis, AI Pixel-Level Manipulation and AI Content Compliance.
Your policy decides what happens next: approve, step up, route to manual review or decline. NeuroVerify does not approve or reject the applicant or the account.
How fast is it, and how is it benchmarked?
Latency is under 200 ms p95 on eligible image requests under defined test conditions. The methodology is available on request, so you can compare it with your own traffic.
On the public VendorBench-100 benchmark, NeuroVerify ranked first of 36 systems, with a detection score (MCC) of 0.876 against 0.646 for the next-best vendor.
Where does the media go, and what deployment options exist?
Deployment options are a managed cloud API, private cloud or customer VPC, a secure container, on-premises, and air-gapped. Pick the one that matches your data residency and security requirements.
The platform holds ISO/IEC 27001, ISO/IEC 27701, ISO/IEC 42001 and SOC 2 Type II certifications. For handling details, see product data handling and the trust centre.
Can we test it on our own media before committing?
Yes. Send a representative sample of confirmed synthetic cases, genuine customers and borderline applications and compare the scores with your own labels. Include genuine captures from low-end handsets and poor lighting so you can see how thresholds behave on real users, not just on attacks.
Contact the team to set up an evaluation. See also how to evaluate a detection API.
Can it detect a synthetic identity if the face is real?
Not by itself. If a real person lends their genuine face to an invented persona, the media is authentic. NeuroVerify then reports on authenticity, and other signals such as device, behaviour and data consistency need to catch the rest.
Evaluate NeuroVerify
Test NeuroVerify on your synthetic identity cases.
Send NeuroVerify a sample of confirmed synthetic and genuine onboarding media and see how the scores separate them.