Notifications
Clear all
Topic starter
19/08/2026 4:05 pm
TL;DR: Financial institutions are shifting model risk management from pre-deployment review to continuous monitoring as machine learning models, changing data, and production drift create new failure modes, according to Fiddler. The practical lesson is that explainability, validation, and lifecycle controls matter more than model performance alone, especially where decisions affect customers, fraud, and regulatory exposure.
NHIMG editorial — based on content published by Fiddler: AI in Finance Panel: Accelerating AI Risk Mitigation with XAI and Continuous Monitoring
By the numbers:
- 30% of one bank model risk team’s models are now machine learning models rather than traditional statistical approaches.
- Fraud models can shift rapidly when customer behaviour changes, as they did in early 2020 when card-not-present transactions surged.
Questions worth separating out
Q: How should financial institutions govern machine learning models after deployment?
A: They should treat deployment as the start of governance, not the end.
Q: Why do explainability and interpretability matter in model risk management?
A: They matter because teams cannot defend, audit, or safely challenge a model they cannot explain.
Q: What breaks when organisations rely only on pre-launch model testing?
A: Pre-launch testing fails when real-world data shifts after deployment, which is common in finance.
Practitioner guidance
- Define approval boundaries for every production model Document the exact use case, decision scope, and business conditions under which each model may operate, then refuse production approval if those boundaries are unclear.
- Separate model build and challenge authority Give independent reviewers authority over validation, sign-off, and escalation so developers cannot both create and certify the same model.
- Instrument continuous drift monitoring Track input drift, output stability, and error spikes, and link them to business thresholds that trigger review before customer harm accumulates.
What's in the full article
Fiddler's full blog covers the operational detail this post intentionally leaves for the source:
- The panel discussion nuances around model risk management operating models and reviewer separation.
- Speaker commentary on how banks are adapting model validation as ML replaces statistical approaches.
- The full context behind continuous monitoring examples, including fraud drift during COVID-era behaviour shifts.
- The broader discussion on explainability, fairness, and responsible AI culture in financial services.
👉 Read Fiddler's panel summary on AI risk mitigation in finance with XAI and continuous monitoring →
AI in finance governance: are explainability and monitoring keeping up?
Explore further
19/08/2026 4:39 pm
AI governance debt is accumulating faster than most finance teams can absorb it: the panel shows that organisations are adding machine learning faster than they are hardening the review, validation, and monitoring processes around it. That creates a backlog of model decisions whose risks are understood only after deployment. The practical conclusion is that governance has to scale with model complexity, not follow it.
A question worth separating out:
Q: Who is accountable when an AI system makes a harmful decision?
A: Accountability should follow the identity chain that authorized, configured, or triggered the action, including the human owner, the platform team, and any delegated agent or tool account. If the organisation cannot name that chain, the governance model is too weak for regulated AI use.
👉 Read our full editorial: AI risk mitigation in finance now depends on explainability and monitoring
