Model Drift
The degradation of a model's predictive performance over time due to changes in the underlying data distribution or relationships between features and the target variable.
In plain language
When an AI's performance slowly declines because the real world has changed. A fraud detector trained on 2020 data may miss new fraud schemes that emerged by 2024.
Why this matters
Model drift is a critical operational risk that governance frameworks must address. Without ongoing monitoring, AI systems degrade silently while appearing to work, potentially making unsafe decisions.
Relevance
ImplementationManaging drift requires continuous monitoring and defined retraining triggers, both of which must be specified in governance policies for each deployed model.
Related terms
Putting model drift into practice in your organisation?
Ready to transform your AI strategy?
Partner with Australia's AI strategy and governance specialists. From adoption roadmaps to ISO 42001 audit readiness.