AjaCertX
ONE ECOSYSTEM. INFINITE STANDARDS.

We Value Your Privacy

We use essential cookies to make our website work. With your consent, we also use optional analytics cookies to understand how visitors use our site.

Cookie Policy  ·  Privacy Policy

You must make a selection to access this website.

Your preference will be saved for future visits.

AI Validation · Monitoring

What Is AI Model Drift

Data drift vs concept drift — why your validation programme must account for it.

October 2025 · 4 min read

A validated AI system can fail without any code changes. This is model drift — gradual performance degradation as the real-world environment diverges from the training environment. EU GMP Annex 22 and the GAMP AI Guide both require ongoing monitoring programmes to detect it.

Data Drift vs Concept Drift

Data drift occurs when the statistical characteristics of input data change — a new raw material supplier, a new manufacturing line, a different patient population. Concept drift occurs when the relationship between inputs and outputs changes — the learned pattern is no longer valid. Both can invalidate a previously validated model without any code changes.

What Your Monitoring Programme Needs

Defined performance KPIs with alert thresholds, monitoring frequency proportionate to risk, a procedure for investigating alerts, and clear revalidation triggers — documented in a monitoring plan as part of the six-document AI inspection package.

Download Whitepaper

Ready to Set the Standard?

Partner with AjaCertX for integrated compliance and assurance solutions.