MacrosLM
Audit & Controls
KYC RISK SCORING / FATF + EU AI ACT / CASE: JPMORGAN / ALSAFA
KYC TEARSHEET · CUSTOMER RISK RATING · FATF / EU AI ACT · Q2 2026
JPMorgan KYC engine · risk-scoring a high-net-worth onboarding
FATF 4-dimension framework applied to a single onboarded entity: ALSAFA Holdings SARL (Luxembourg). Composite risk 78/100 — High. Triggers EDD, beneficial-owner trace, and 90-day pKYC cycle.
INSTITUTION JPMORGAN PRIVATE BANK · CUSTOMER ALSAFA HOLDINGS SARL · FRAMEWORK FATF + EU AI ACT (AUG 2026) · ENGINE MacrosLM KYC
Composite risk score
78/100
High — tier 3 of 4
Required posture
EDD
Enhanced due diligence
pKYC review cadence
90d
vs std 365d
Explainability (EU AI Act)
Pass
Score + reasons logged
01FATF 4 DIMENSIONS
02CUSTOMER SCORE
03TIER → EDD MAPPING
04PERPETUAL KYC
05EU AI ACT
§ 01The FATF four-dimension framework
FATF Recommendation 10 requires risk-based KYC. The four dimensions — customer, geography, product, and channel — combine into a single composite score. JPMorgan applies each dimension with its own factor library and weights tuned annually against actual SAR-filing outcomes.
Engine parameters
- Framework
- FATF Rec 10 + FFIEC BSA/AML manual
- Engine
- MacrosLM KYC Risk Scoring
- Scoring scale
- 0 – 100 composite
- Tiers
- 0–24 Low / 25–49 Medium / 50–79 High / 80+ Prohibited
- Refresh cycle
- Event-triggered + scheduled (pKYC)
- Last calibration
- Mar 2026 · against 38K closed cases
- Reg lens
- EU AI Act · high-risk system
Dimension weights · JPM Private Bank · 2026 calibration
| Dimension | Sub-factor count | Weight | Top drivers |
|---|
| 1. Customer | 17 sub-factors | 35% | PEP status, beneficial-owner opacity, source of wealth |
| 2. Geography | 12 sub-factors | 25% | FATF grey/black list, sanctions exposure, residency vs nationality split |
| 3. Product | 14 sub-factors | 25% | Cash intensity, cross-border flow, trust/SPV structures, crypto rails |
| 4. Channel | 9 sub-factors | 15% | Non-face-to-face, third-party introduction, digital-only onboarding |
| Total | 52 | 100% | Composite via weighted sum + adjustment rules |
§ 02Customer scoring · ALSAFA Holdings SARL
Fictional onboarding case. Luxembourg holding entity, beneficial owner family office in MENA, wealth from real-estate inheritance. The case illustrates a textbook High-tier classification — no single fatal factor, but multiple moderate-elevated indicators combine into a score that crosses the EDD threshold.
Per-dimension breakdown · ALSAFA Holdings SARL
| Dimension | Sub-factor | Customer attribute | Factor score | Weighted | Pill |
|---|
| Customer | Entity type | Luxembourg SARL holding co. | 60 | — | Elevated |
| PEP status | UBO is family member of regional minister | 90 | — | High |
| Beneficial ownership | 3-layer structure · UBO trace possible | 55 | — | Elevated |
| Source of wealth | Inherited real-estate · independently verified | 30 | — | Acceptable |
| Subtotal × 35% | — | 68 | 23.8 | — |
| Geography | Country of incorporation | Luxembourg · FATF compliant | 15 | — | Low |
| UBO residency | MENA jurisdiction · grey-list adjacent | 75 | — | Elevated |
| Wealth-source jurisdiction | Same as UBO residency | 70 | — | Elevated |
| Subtotal × 25% | — | 62 | 15.5 | — |
| Product | Account type | Private bank multi-currency + custody | 55 | — | Elevated |
| Expected flow | $80M initial · $5–15M monthly | 75 | — | Elevated |
| Cross-border | Predominantly intra-EU + UAE | 65 | — | Elevated |
| Subtotal × 25% | — | 65 | 16.3 | — |
| Channel | Onboarding mode | In-person via Lux RM + Geneva visit | 25 | — | Acceptable |
| Introduction | Existing JPM client referral | 20 | — | Acceptable |
| Subtotal × 15% | — | 22 | 3.3 | — |
| Composite | Weighted sum + 5 pt PEP adj. | — | — | 78 | High · tier 3 |