Which of the following controls BEST mitigates the risk of data poisoning?

The correct answer is B. Data validation. Data poisoning attacks inject malicious, corrupted, or manipulated data into training or input datasets to distort model behavior. Data validation (B) is the most direct preventive control - it verifies the integrity, source, format, and legitimacy of data before it is accepted…

AI Security Design and Implementation

Question

Options

AData set restoration
BData validation
CDigital watermarking
DIntrusion detection

How the community answered

(31 responses)

A
3% (1)
B
94% (29)
D
3% (1)

Explanation

Data poisoning attacks inject malicious, corrupted, or manipulated data into training or input datasets to distort model behavior. Data validation (B) is the most direct preventive control - it verifies the integrity, source, format, and legitimacy of data before it is accepted into the system, blocking poisoned data before it can influence the model. Dataset restoration (A) is a recovery control that addresses consequences after poisoning occurs. Digital watermarking (C) helps trace data provenance but does not prevent malicious data from being introduced. Intrusion detection (D) may detect anomalous activity but does not validate the content of data itself. Only data validation acts as a gate against poisoned inputs.

Topics

#Data poisoning#Data validation#AI security controls#Machine learning security

Community Discussion

No community discussion yet for this question.

Full AAISM Practice