Your image
Rescaled to 500 px on the longest side. Discarded after analysis.
AppCraft image-forensics · proprietary stack · ~1 s
Is this image
real?
Is this image
real?
Drop it in. We’ll tell you.
A live demo of the image-forensics service AppCraft deploys for banks, insurers and KYC platforms. We combine pixel-level Error Level Analysis with our own detection stack to flag content from every major image generator in production today — and to catch the human-edited cases AI classifiers miss.
Your image is processed and discarded
No signup, no tracking pixel
Sub-second per image
Technical screener, not a verdict. This is an informational forensics check — not proof an image is real or fake. A skilled fake can pass and a genuine photo can trip a flag, so treat the result as a signal to verify, not a decision.
Privacy: your image is processed only to run the check and auto-deleted within 7 days. We never train models on it. See our privacy & data policy.
How to think about a suspicious image
A detector is a starting point, not the answer. The habit matters more than the tool:
- Pause. A result — ours or anyone’s — is a signal, not a verdict. Don’t act on it alone.
- Ask for corroborating facts. Where did this image come from? Is there a second, independent source? Who benefits if you believe it?
- Learn the patterns. Most synthetic fraud reuses the same tricks — mismatched lighting, impossible details, urgency that rushes you. Once you know them, you carry the skill even when the tools change.
Coverage
Detects content from every major image generator.
Our stack updates monthly to keep up with the model release cadence. You don’t pin a version of an AI detector — we do.
- MidJourney
- OpenAI DALL·E
- OpenAI Sora
- Google Gemini / Imagen
- Anthropic Claude
- Stable Diffusion
- Adobe Firefly
- Black Forest Flux
- Ideogram
- Recraft
- Leonardo.ai
- Runway Gen-4
- Kling
- and many others
From our case work
What our stack catches in the wild.
Five representative examples from real banking, insurance and KYC workflows. The two side-by-side pairs let you compare an authentic capture with the AI or deepfake equivalent. The three single-image cards are AI-generated submissions we ran through our own forensics pipeline.
Banking onboarding
Synthetic portrait detection
Side-by-side: an authentic user-supplied photo (left) vs an AI-generated face (right) submitted to a video-KYC flow. Our stack flags the second one in under a second.
Identity verification
Face-swap deepfakes
Same person on the left, a celebrity face swapped onto a real photo on the right. The donor-to-host pixel boundary gives the manipulation away.
Auto insurance
Manipulated · 96%
Object inserted into a real frame
A genuine vehicle photo with a second car composited in. Our pixel-history map highlights exactly where the added region sits — bottom-right of the frame in this case.
Property insurance
AI-generated · 88%
Burned-boiler claim photo
An AI-fabricated photo of a destroyed gas water heater, submitted as evidence of damage that never happened. No camera trail, anomalous pixel statistics across the charred regions.
KYC fraud
AI-generated · 96%
Synthetic South African Smart ID
An AI-fabricated South African Smart ID Card — the exact format SA banks and fintechs verify every day in onboarding. Visually indistinguishable from a real card at a glance, yet three of our four channels fire on it within a second.
The boiler and ID-card images above were generated by us (OpenAI gpt-image-2, May 2026) specifically to demonstrate detection of synthetic claim photos and synthetic documents — two of the highest-volume attack categories we see across African banking and insurance.
Why this matters now
Africa’s image-fraud crisis, in 2026 numbers.
Generative AI dropped the cost of fabricating a believable claim photo, passport scan or KYC selfie to near zero. The attack volume followed — and Africa is feeling it first.
87%
AI-driven biometric fraud
Of rejected biometric verifications in Southern Africa are linked to AI-assisted impersonation.
Smile ID Digital Identity Fraud Report 2026
69%
African fintech fraud is AI-generated
Share of biometric fraud cases tracked across African fintechs that involve generative AI.
Smile ID 2026
167%
Surge in identity fraud
Year-on-year increase in identity-fraud incidents on the continent, with deepfakes leading the trend.
Sumsub Identity Fraud Report
100k+
Injection attacks per month
Emulator, virtual-camera and screen-recording attacks targeting KYC flows — per month, across Africa.
Smile ID 2025
$32M
Nigerian bank fraud losses
Annual fraud losses in Nigerian banking (₦52.26B), up 196% over five years.
NIBSS / Businessday 2025
30%
Of enterprises hit by deepfake KYC by year-end
Global Gartner forecast for 2026 — and Africa is moving faster than the global average.
Gartner / Jumio 2026
Sources: Smile ID Digital Identity Fraud Report 2026, Sumsub Identity Fraud Report 2025, NIBSS / Businessday Nigeria 2025, Gartner / Jumio 2026, Insurance Crime Bureau (South Africa) 2025.
Built for
Where this lives in your stack.
We deploy AppCraft image-forensics as an on-prem service, in your VPC, or as a managed REST/gRPC API behind your own ingress. It plugs into the workflows your team already uses.
KYC onboarding
Catch AI-generated faces and document scans before they enter your verification queue. Sub-second per image, audit-trail logging your compliance team will accept.
Claim triage
Route claim photos through forensics before they hit the adjuster’s desk. Flag AI-fabricated damage and edited receipts up front instead of in post-payout audits.
Newsroom & e-services
Vet user-submitted photos for editorial publication, or screen citizen-portal uploads (permit photos, document scans) for synthetic content at scale.
Need this in production?
API access, bulk verification and on-prem deployment for medium and large businesses.
Tell us about your volume and your verification flow. We’ll come back with a sandbox key, an integration plan and a price.