Artificial Intelligence Governance Professional AIGP · Free Practice Question Hard
Question 35
A company deploys a facial recognition security system at building entrances. Attackers use specially designed glasses that fool the AI into misidentifying unauthorized individuals as employees. Which defense would MOST effectively reduce this risk?
- A Limit model training to only authorized employee faces
- B Increase image resolution only
- C Implement regular model retraining with diverse, real-world adversarial samples
- D Rely solely on manual badge checks
Reveal correct answer
Correct answer: C
Explanation
Training on adversarial examples increases model resilience to such attacks and is a recommended best practice.A. This limits scope but doesn't address adversarial manipulations targeting the model itself.
B. While higher resolution may help, it does not directly address adversarial manipulation.
C. This defense helps the model recognize and resist such deceptive patterns in the future.
D. Manual checks reduce reliance on AI but don't improve model robustness or security.
Discussion
Think the marked answer is wrong, or have a better explanation? Share it below — comments appear after review.
You must be logged in to post a comment.
