aml-typologies

name	aml-typologies
description	Activate for: AML alert, transaction monitoring, suspicious activity, money laundering, structuring, smurfing, round-tripping, layering, placement, integration, typology, red flag, velocity alert, geographic anomaly, TBML, trade-based money laundering, network analysis, PEP alert. NOT for: SAR/STR drafting or filing (use aml-sar-drafting), customer onboarding or KYC documentation (use aml-cdd-edd), sanctions list screening (use sanctions-screening).
metadata	{"version":"1.0","author":"Panaversity — The AI Agent Factory","standard":"FATF 40 Recommendations — jurisdiction AML legislation (load overlay)"}

THE THREE STAGES OF MONEY LAUNDERING

Placement — Introducing illicit funds into the financial system Red flags: Large cash deposits; multiple deposits just below reporting threshold (structuring); money service business transactions; casino chip purchases/redemptions.

Layering — Obscuring the audit trail through multiple transactions Red flags: Rapid movement through multiple accounts; complex wire sequences across multiple jurisdictions; shell company or nominee director use; conversion between asset classes (cash → wire → crypto → real estate).

Integration — Re-introducing laundered funds as apparently legitimate income Red flags: Luxury asset purchases (real estate, vehicles, art); unexplained business investments; loan repayments from obscure sources; complex "business transactions" with no apparent commercial purpose.

TOP 20 AML TYPOLOGIES WITH DETECTION LOGIC

1. STRUCTURING (SMURFING) Indicator: Multiple cash deposits/withdrawals in a single account, each just below the reporting threshold, within a defined period. Detection rule: ≥3 transactions within 7 days each ≥ 80% of the reporting threshold. ML enhancement: Aggregate transactions per customer per rolling 30-day window. Note: The reporting threshold is jurisdiction-specific (USD 10,000 in USA, varies elsewhere).

2. ROUND-TRIPPING Indicator: Funds sent from Account A to Jurisdiction X, then returned from Jurisdiction X to Account A or a closely connected account within a short period. Detection: Wire-out to jurisdiction X followed by wire-in from same jurisdiction within 30 days, same approximate amount (±10%).

3. RAPID FUND MOVEMENT / VELOCITY Indicator: Account receives large funds then transfers them out within 24–72 hours with minimal balance remaining. Detection rule: Opening balance < 10% of daily inflow; >80% of incoming funds transferred out within 48 hours.

4. GEOGRAPHIC ANOMALY Indicator: Transactions to/from high-risk jurisdictions inconsistent with customer profile. Detection: Any transaction above de minimis to FATF grey-listed or black-listed countries. Enhanced scrutiny for: Iran, North Korea, Myanmar (FATF black list); current grey list — check FATF.org for current listing as it changes quarterly.

5. TRADE-BASED MONEY LAUNDERING (TBML) Indicator: Over/under-invoicing; multiple invoicing; falsely described goods. Red flags: Invoice amounts inconsistent with commodity market price; import/export volumes inconsistent with declared business size; inconsistency between shipping documents and invoices.

6. PROFESSIONAL MONEY LAUNDERING NETWORKS Indicator: Account connected through transactions to other accounts with known AML suspicion or law enforcement interest. Detection: Graph network analysis — identify high-centrality nodes, clusters of accounts with high internal velocity and cash-out at network periphery.

7. PEP UNUSUAL TRANSACTION Indicator: Politically Exposed Person receives unexplained large incoming transfer disproportionate to known income/wealth profile. Detection rule: Any PEP account single transaction > 3× average monthly balance. Any PEP incoming transfer from jurisdiction with high corruption index.

8. SHELL COMPANY LAYERING Indicator: Corporate account with no apparent operational activity — no payroll, no supplier payments, only wire receipts and outflows. Red flags: Beneficial owner structure through Cayman / BVI / Panama / Cyprus; multiple directors with no apparent operational role; no website or publicly verifiable business presence.

9. LOAN-BACK Indicator: Customer places cash/funds in account, then takes a "loan" collateralised by those funds. Repays loan with laundered funds; withdraws original "collateral" clean. Detection: Secured lending against own deposits; loan repayments from sources inconsistent with declared income.

10. REAL ESTATE MONEY LAUNDERING Indicator: All-cash property purchase; multiple rapid buy/sell transactions in same property; purchase at significantly above or below market value. Red flags: Payment from a third party with no apparent connection to the buyer; offshore corporate as buyer with opaque beneficial ownership.

11. CRYPTO-TO-FIAT CONVERSION Indicator: Incoming wires from cryptocurrency exchanges to bank accounts. Red flags: High-frequency conversion from multiple exchanges; exchanges operating in high-risk jurisdictions with weak AML controls; transaction amounts not rounded (characteristic of crypto exchange deposits).

12. MONEY MULE ACTIVITY Indicator: Account receives funds from multiple unknown sources and immediately transfers to third parties; account holder's profile inconsistent with transaction volumes. Detection: Multiple incoming wires from unknown parties; average transaction ≫ declared income; account holder has changed personal details (address, employment) recently.

13. INSURANCE PREMIUM FINANCING LAUNDERING Indicator: Purchase large-value life insurance policy with lump sum; cancel early and accept surrender penalty to receive "clean" refund cheque from insurer. Red flags: Large single premium life insurance; early policy surrender; cash source inconsistent with declared income.

14. INVOICE FRAUD / ADVANCE FEE FRAUD Indicator: Account receives multiple large payments described as "invoices" or "commissions" from offshore entities with no apparent business relationship.

15. LAYERING VIA SECURITIES Indicator: Purchase securities (equities, bonds) with cash/wire, then rapidly sell and withdraw cash from proceeds. Detection: Same-day or next-day securities purchase and sale with no apparent investment rationale; securities transactions from accounts with no investment history.

16. CROSS-BORDER CASH SMUGGLING FRONT Indicator: Business with declared import/export activities used to justify cross-border fund movements that are disproportionate to the declared business scale.

17. HAWALA / INFORMAL VALUE TRANSFER Indicator: Payments to known hawala operators; international transfers with "settlement" description; amounts consistent with remittance fees charged on informal value transfer.

18. GAMBLING / CASINO LAYERING Indicator: Purchase casino chips with cash; minimal gambling; cash out. Remote gambling: Fund online gambling account; withdraw without significant play.

19. CORRESPONDENT BANKING NESTED ACCOUNTS Indicator: Respondent bank (foreign) maintains accounts for its own customers within the correspondent relationship — "nested" accounts not directly visible to the correspondent bank. Detection: Wire patterns suggesting multiple underlying account holders using a single correspondent relationship.

20. TERRORIST FINANCING (SMALL AMOUNTS, UNUSUAL DESTINATIONS) Indicator: Multiple small international transfers to individuals in conflict zones or areas with known terrorist activity. Distinguishing feature: Unlike money laundering (large amounts laundered to appear clean), terrorist financing often involves small amounts from legitimate sources.

ALERT DISPOSITION DECISION TREE

Level 1 — Quick Close (< 5 min): Fully explained by known business profile? YES with documentation → Close with documented rationale

Level 2 — Customer/RM Contact (< 1 hr): Plausible explanation but insufficient info? Request clarification → If satisfactory + documented → Close If unsatisfactory → Escalate to Level 3

Level 3 — Enhanced Investigation (< 24 hr): Suspicious despite explanation? Obtain: company filings, adverse media search, transaction history, BO verification If suspicion eliminated + documented → Close If suspicion remains → Escalate to Level 4

Level 4 — MLRO SAR Decision: Apply jurisdiction's SAR standard (see overlay) UK: "reasonable grounds to suspect" (POCA 2002) USA: "reason to suspect" involving $5,000+ and a legal violation (BSA) File SAR/STR if standard met → Account continuation decision

MACHINE LEARNING ENHANCEMENTS

Peer group analysis: Flag customers whose transaction patterns deviate >2 standard deviations from statistically similar peers (same product, geography, business type). Network/graph analysis: Flag customers who are highly connected through transactions to other flagged/suspicious customers — even if their own transactions appear benign. Autoencoder anomaly detection: Train on "normal" transactions; flag those with high reconstruction error as novel — effective for detecting new typologies. Reported outcomes: Featurespace/Visa and NICE Actimize report 50–70% false positive reduction in production deployments while maintaining or improving detection rates.

OUTPUT FORMAT — TYPOLOGY ASSESSMENT

TYPOLOGY ASSESSMENT
Alert ID:          [System-generated alert ID]
Customer:          [Name / ID]
Assessment Date:   [YYYY-MM-DD]

ALERT TRIGGER:
  Rule/Model:      [Rule name or ML model that generated the alert]
  Transaction(s):  [Date, amount, type, counterparty for each flagged transaction]

TYPOLOGY IDENTIFIED:
  Primary:         [Typology name from list above — e.g., STRUCTURING, ROUND-TRIPPING]
  Secondary:       [Additional typology if applicable]
  ML Stage:        [Placement / Layering / Integration]

ASSESSMENT:
  Customer Profile:    [Summary of expected activity vs. observed activity]
  Innocent Explanation:[Considered and why insufficient, OR why activity is legitimate]
  Risk Indicators:     [List specific red flags observed]

DISPOSITION:
  Level:           [1 — Quick Close / 2 — RM Contact / 3 — Enhanced Investigation / 4 — MLRO SAR Decision]
  Decision:        [Close with rationale / Escalate / File SAR]
  Rationale:       [Documented reasoning for disposition]

NEVER DO THESE

• NEVER close an alert without documenting the rationale — undocumented closures are treated as control failures by regulators and auditors
• NEVER apply a single typology's detection rule in isolation without considering whether multiple typologies may be operating simultaneously (e.g., structuring + round-tripping)
• NEVER dismiss geographic anomaly alerts solely because the customer has a stated business reason for transacting with high-risk jurisdictions — the business explanation must be independently verified against trade documents and commercial reality
• NEVER use ML model output alone as the sole basis for closing or escalating — model scores inform human judgment but do not replace the analyst's documented assessment
• NEVER treat terrorist financing alerts using the same thresholds as money laundering alerts — TF often involves small amounts from legitimate sources directed to conflict zones

ALL OUTPUTS REQUIRE REVIEW BY A QUALIFIED PROFESSIONAL BEFORE USE IN REGULATORY FILINGS OR BUSINESS DECISIONS.