AAISM Question 16

Community Discussion

Written by other learners. Comments can disagree with the marked answer — check the official documentation when in doubt.

• Chose A2

  Physical alterations in facial appearance (such as makeup, disguises, accessories, or adversarial patterns) are designed to trick or evade facial recognition systems (evasion attacks). By incorporating a diverse set of augmented training data—spanning different angles, lighting conditions, expressions, and simulated alterations—you harden the core model's robustness, making it significantly less susceptible to being fooled.

• 1

  The attacker deliberately alters facial appearance to cause incorrect recognition. This is a form of model evasion, as described in the ISACA Official Review Manual, §2.7.1. increasing training-data variability improves the facial recognition model's ability to recognise individuals despite alterations in their appearance, thereby reducing susceptibility to model-evasion attacks. Answer D is questionable and depends on the secondary system security

• Chose A1

  At first glance, D sounds attractive: Implementing a secondary AI system to confirm images. This is a compensating control, not a root-cause solution.