Evidence
Official pilot API surface
POST /v1/bankingx40/qeiv15/aml-kyc-triage
Controlled endpoint run
| Evidence item | Result |
|---|---|
| Controlled AML/KYC cases | 100 |
| Repeats per case | 3 |
| Official endpoint runs | 300 |
| QEIv15 changed decisions | 30 |
| RELEASE → REQUIRE_VERIFICATION | 15 |
| REQUIRE_VERIFICATION → BLOCK | 15 |
| Repeatability checks | Passed |
| Anti shortcut checks | Passed after repair and rerun |
| Independent verification | Passed |
Technical trust note
During independent freeze archive review, a hardcoded proof case marker was identified, removed, and the review was rerun successfully. The repaired review passed with zero failures.
What this proves
BankingX40 has controlled official endpoint evidence that deterministic contract rules plus QEIv15 structural boundary metrics can alter AML/KYC AI output decisions in defined boundary cases.
What this evidence is for
This evidence supports controlled bank or RegTech review using synthetic, sanitized, or approved non production cases.