Autonomous Red-Team / Blue-Team Payment Fraud Simulation & Closed-Loop Defense Engine · Mastercard AI Defence Lab (GFF 2026) · Submitted by Kanak Sanjay Waradkar
Stop card testing, bot traffic, and fake signups on any site. Device identity that survives cleared cookies and rotating IPs, velocity across device/subnet/ASN/BIN, and a reason code behind every verdict. Drop-in script tag, self-hosted, MIT.
Block fake orders in WooCommerce
Pre-authorization behavioral risk system that detects automated card-testing sequences before a Razorpay payment order is created.
Deterministic streaming detector for card-testing and velocity abuse on card authorization traffic. Rust core + bit-exact Python reference, temporal-split evaluation, false-positive cost in rupees. Precision 0.0824 at a realistic base rate; declines 1 in 71 legitimate customers. Not deployable as-is.
Merchant-side card-testing detection that shows its reasoning. Rules produce evidence, five explanations compete over it, and the model can ask for a review but never blocks a shopper — Razorpay Buildathon 2026, Track 02.
To associate your repository with the card-testing topic, visit your repo's landing page and select "manage topics."