Which of the following is the MOST likely cause of model drift?

The correct answer is A. Data poisoning. Model drift occurs when the statistical properties of input data and/or the relationship between features and outcomes change over time, causing degraded model performance. The AAISM guidance classifies data-centric causes (distribution shift, concept drift, and contamination) as

AI Security Risk Management

Question

Options

AData poisoning
BPerfect knowledge
CMembership inference
DModel stealing

How the community answered

(51 responses)

A
94% (48)
C
4% (2)
D
2% (1)

Explanation

Model drift occurs when the statistical properties of input data and/or the relationship between features and outcomes change over time, causing degraded model performance. The AAISM guidance classifies data-centric causes (distribution shift, concept drift, and contamination) as the primary drivers and highlights that malicious contamination of training or incremental learning data (data poisoning) is a direct, high-likelihood driver of observable drift in production because it changes the effective data-generating process the model learns from.

Topics

#Model Drift#Data Poisoning#Adversarial Attacks#AI Model Reliability

Community Discussion

No community discussion yet for this question.

Full AAISM Practice