Ikibazo gihuza ibibazo · Hagati Urwego

Ikigereranyo cyabakiriya muri AML Ikibazo

Sobanukirwa n’ingaruka zishingiye ku bakiriya ba AML, ibiciro bya ML, imiterere, imipaka yerekana imipaka, ubutabera, isuzuma rihoraho hamwe n’ibyemezo byabantu.

Inzira zijyanye no kuyoboraIkigereranyo cyabakiriya Muri Aml
Ikibazo 1 Bya 8

How does a risk rating support AML customer due diligence?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. How does a risk rating support AML customer due diligence?
    • Proving that a customer committed a crime
    • Assigning one uniform level to every person in a broad category
    • Organizing due-diligence and monitoring effort
    • Replacing customer information with a single score
  2. Why should a bank avoid assigning one risk level to every customer of a type?
    • Customer type contains all required facts
    • Risk depends on facts and circumstances of the individual relationship
    • Risk ratings are never allowed
    • The model automatically detects every exception
  3. How can machine learning support customer risk review?
    • Automatically determine criminal intent
    • Remove the need for due diligence
    • Guarantee no suspicious activity is missed
    • Prioritize patterns or changes for analyst attention
  4. How can historical suspicious-activity labels limit model generalization?
    • They reflect only activity previously detected and reported
    • They include every instance of illicit activity
    • They are always immediate and complete
    • They are unaffected by institutional policy
  5. When might a customer risk profile need review?
    • Only at account closure
    • Whenever a model score is low
    • When ownership, products, geography or behavior materially changes
    • Never after onboarding
  6. Why monitor override and closure rates across populations?
    • To make every rating identical
    • To detect inconsistent or disproportionate outcomes
    • To increase transaction volume
    • To avoid reviewing model inputs
  7. What should a high model score normally trigger?
    • Automatic public accusation
    • Guaranteed account closure
    • Deletion of customer records
    • Proportionate review under documented policy
  8. Who remains accountable for applying AML requirements?
    • The institution's responsible personnel and compliance processes
    • The model provider alone, after deployment
    • The training data, because it defined the labels
    • The customer whose account received the score