For Heads-of · Practitioner
AI incident response and rollback
Defined containment, remediation, and rollback procedure for a detected AI system failure or harmful output.
- corrective
- incident-response
- remediation
What it does
Defines the containment, remediation, and rollback procedure to run when an AI system produces a harmful or non-conforming output in production.
Where it fits
The corrective closing loop after monitoring detects a failure — without it, detection doesn't translate into a fixed system.
Risks this mitigates
The risks this control addresses, ranked by effectiveness.
Training data bias and discriminatory outcomes
Historical or sampling bias in training data is reproduced or amplified in model outputs and decisions.
Third-party model and dataset provenance
A model, dataset, or plugin is adopted from an external source without verifying provenance, licensing, or update history.
AI model drift and degradation
A deployed model's performance silently degrades over time as the input distribution shifts away from its training and validation data.
Untested AI failure scenarios
No scenario testing exists for AI-specific disruption modes against the organisation's impact tolerances.