AI0-001 AI Security Practice Question
A data scientist wants to protect the privacy of individuals whose data is used to train a model, even if the model is compromised. Which technique ensures that the model does not memorize sensitive information?
⚠ Common exam trap
CompTIA often tests the misconception that data anonymization (D) is sufficient for model privacy, but candidates must recognize that anonymization does not protect against model inversion or membership inference attacks, whereas differential privacy provides a formal mathematical guarantee.
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
✓
Differential privacy
Differential privacy (C) is the correct technique because it adds calibrated noise to the training data or model updates, ensuring that the model's outputs do not reveal whether any specific individual's data was included. This guarantees that even if an attacker gains full access to the model, they cannot extract sensitive information about any single record, as the noise bounds the influence of any one data point.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Federated learning
Why it's wrong here
Federated learning keeps raw data on local devices and shares only model updates, so it addresses data-centralisation during training, not memorisation. A compromised model can still leak training records. It is the right choice when data cannot leave edge devices, such as hospital silos, but it does not guarantee non-memorisation.
- ✗
Homomorphic encryption
Why it's wrong here
Homomorphic encryption permits computation on ciphertext, yet training still exposes gradients derived from plaintext records, so memorisation persists. It suits inference over encrypted inputs where the server must never decrypt them. The scenario instead requires differential privacy, which bounds any single record's influence on the trained model.
- ✓
Differential privacy
Why this is correct
Differential privacy injects calibrated noise into training or query outputs, bounding any single individual's influence so the model cannot memorise their record. This satisfies the requirement that privacy survives model compromise, unlike encryption or anonymisation, which protect data at rest rather than learned parameters.
- ✗
Data anonymization
Why it's wrong here
Anonymisation strips direct identifiers before training, but a compromised model can still memorise and regurgitate latent patterns from the underlying records, so it does not guarantee non-memorisation. It is tempting because it is the standard governance control for sharing or publishing datasets, and would be correct when the goal is de-identifying stored data rather than hardening a trained model.
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.