Generative AI Leader · Free Practice Question Medium
Question 19
To prevent unauthorized modifications or tampering of a deployed generative AI model that provides financial advice, a company implements measures to verify the model's integrity before each use and protect its stored artifacts.
This focus on protecting the model itself from corruption or malicious alteration is a key aspect of:
-
A
Securing the AI model and its assets
-
B
Data privacy for training inputs
-
C
Ensuring model fairness and mitigating bias
-
D
Optimizing model inference latency
Reveal correct answer
Correct answer: A
Explanation
Protecting the model from unauthorized changes ensures its integrity, a core part of AI security.
INCORRECT: Data privacy for training inputs
This concerns the data used to train the model, not the security of the deployed model artifact.INCORRECT: Ensuring model fairness and mitigating bias
This relates to responsible AI principles in model behavior, not directly to protecting the model artifact from tampering.CORRECT: Securing the AI model and its assets
Securing AI involves protecting all components of the AI system, including the trained model itself. Measures to ensure model integrity, protect against unauthorized access or modification of model artifacts, and verify a model's authenticity before use are crucial for preventing malicious actors from corrupting the model or causing it to behave in unintended, harmful ways.INCORRECT: Optimizing model inference latency
This is a performance consideration, not a security measure for model integrity.
Discussion
Think the marked answer is wrong, or have a better explanation? Share it below — comments appear after review.
