An incorrect chatbot answer caused a resident-service incident. After containment, what sustainment step should follow?

After an incorrect chatbot answer caused a resident-service incident and containment is complete, the sustainment step that should follow is a post-incident review that updates AI answer-failure playbooks. Containment stops immediate harm to the resident and the service channel; learning hardens detection, escalation, correction messaging, and prevention so the next miss is handled faster and is less likely to recur. Playbook updates convert a single civic failure into lasting institutional capability for clerks and AI ops. Deleting all logs so the incident cannot be discussed again fails accountability and learning requirements for public organizations and invites repeat harm. Blaming a single clerk and closing the matter without process change fails because it misattributes systemic answer failure to an individual and leaves monitoring, content, and escalation controls unchanged. Disabling monitoring so fewer incidents are detected next time fails by confusing blindness with safety and undermines civic oversight. Exam caveat: blame and log-deletion distractors test ethics and learning culture as much as technical response after AI incidents. Operational check: run a blameless post-incident review, capture timeline and contributing factors, update answer-failure playbooks and monitoring thresholds, assign corrective actions with dates, and brief leadership on residual resident risk.