AI0-001 AI Security Practice Question
An organization deploys a machine learning model for credit scoring. An attacker submits carefully crafted loan applications that are slightly outside normal ranges but cause the model to approve high-risk loans. What type of attack is this?
⚠ Common exam trap
CompTIA often tests the distinction between data poisoning (training-time attack) and adversarial examples (inference-time attack), so candidates mistakenly choose data poisoning when they see 'crafted inputs' without recognizing the attack occurs after deployment.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Adversarial example
This is an adversarial example attack, where the attacker crafts inputs with small, carefully chosen perturbations that cause the ML model to misclassify them. In credit scoring, submitting loan applications with values slightly outside normal ranges exploits the model's decision boundary to approve high-risk loans, a classic evasion technique.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Model extraction
Why it's wrong here
Model extraction reconstructs a copy of the model through repeated queries, stealing intellectual property rather than manipulating outputs. It is tempting because both attacks query the model repeatedly, but the described crafted inputs that force incorrect high-risk approvals are adversarial examples, an evasion attack altering predictions.
- ✗
Prompt injection
Why it's wrong here
Prompt injection manipulates a large language model's instructions through crafted input text, which does not apply to a numeric credit-scoring classifier. It is tempting because both involve maliciously crafted inputs, but prompt injection targets generative LLM behaviour, whereas this scenario describes adversarial examples causing misclassification.
- ✓
Adversarial example
Why this is correct
The attacker perturbs input features slightly so the model misclassifies them, exploiting the decision boundary rather than the training data or model weights. Crafted applications just outside normal ranges that flip approval decisions are the defining signature of an adversarial example.
- ✗
Data poisoning
Why it's wrong here
Data poisoning alters the training dataset to corrupt the model's learned behaviour, but this scenario describes an attacker manipulating inputs at inference time—submitting crafted loan applications—not corrupting historical training data. It is tempting because both attacks exploit model vulnerabilities, and data poisoning would be correct if the attacker had injected malicious records into the credit history dataset to skew approval boundaries during training.
About these practice questions
This AI0-001 question is part of Courseiva's 962-question bank — original exam-style content with full explanations and wrong-answer analysis, never real exam questions or exam dumps. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This AI0-001 practice question is part of Courseiva's free CompTIA certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the AI0-001 exam.