Check a Payment Slip Before You Trust It
Upload a payment screenshot or receipt to examine possible editing, image manipulation, synthetic content, and suspicious transaction details.
SlipShield AI examines the evidence present in an uploaded file. It cannot independently confirm that money has reached your account.
1
Upload your image or PDF
2
Run forensic checks
3
Review the findings and verify the transaction with your payment provider
Edited Image Detection
Payment Detail Extraction
Forensic Risk Report
Upload Payment Slip
JPG, PNG, WEBP or PDF — maximum 15 MB. Your file stays on your device; nothing is uploaded to a server.
or drag & drop the file here
Demo Mode — Educational Samples
Every sample is labelled DEMO — NOT A REAL PAYMENT RECEIPT. Names, accounts and references are fictional.
Analysis
- Reading the uploaded file—
- Extracting image metadata—
- Reading text using OCR—
- Checking image structure and compression—
- Examining potential editing indicators—
- Running available AI-image detection—
- Preparing the report—
Extracted Payment Details (you can correct OCR mistakes)
Your edits are saved as corrections only — the report always shows the original OCR value and its confidence next to your corrected value. Fields that are not in the image show "Not detected". Nothing is carried over from a previous file: every upload starts a new session.
Advanced Analysis
Raw OCR Text
(not run yet)
Forensic Viewer
Highlighted regions require human review. They are NOT automatic proof of fraud.
Optional: Document-Forgery Model Backend (self-hosted, off by default)
The investigated models cannot run in this free static build: zodumair/document-forgery-detector is not publicly accessible on Hugging Face (page/API return 401; weights and license unverified) and Jason37437/SEED is a ~1.26 GB PyTorch model needing a GPU/CPU backend. If YOU run the included backend (backend/ in the source), paste its URL below. The image is sent ONLY when you tick consent and press the button.
Model status: Unavailable (no backend configured). No result is simulated.
Forensic Report
Verify Whether the Money Actually Arrived
Privacy
All OCR and image analysis runs locally in your browser. Files are not sent to any server, not stored, and not used for training.
Implemented: real OCR (Tesseract.js), EXIF/metadata (ExifReader), PDF pages (PDF.js), SHA-256 file hash, ELA, noise / sharpness / duplicate-region analysis — all on-device. Unavailable in this free build: calibrated AI-generated receipt detection and official provider template comparison — the report says so instead of guessing. No accounts, no history, no uploads stored.