Finance notices AI scoring that highlights odd refund chains on business-license payments after hours. Which AI-assisted security task best describes this capability?

AI-assisted fraud detection scores transactional behavior to highlight anomalies such as odd after-hours refund chains on business-license payments, giving finance a prioritized queue for investigation rather than raw ledger noise. In municipal revenue systems, that capability protects license fees and related funds by surfacing patterns consistent with refund abuse, account takeover, or insider manipulation without replacing policy-based payment rules. Access anomaly scoring for late-night licensing portal login events is authentication monitoring; it does not analyze payment anomaly scores or refund chains. Vulnerability ranking for licensing payment portal software flaws is application security triage; it does not score payment behavior or catch refund fraud, and it does not itself protect revenue integrity. Threat intel correlation for licensing database intrusion indicators focuses on external attack signs rather than scoring payment transactions, so it does not surface the refund-chain anomalies finance needs to investigate. Exam caveat: stems that mention AI scoring odd payment or refund behavior point to fraud detection—not access anomaly scoring, vulnerability ranking, or threat intel correlation. Operational check: pair anomaly alerts with immutable audit trails, dual-control refunds on business licenses, and finance review before any automated refund hold becomes a permanent denial of legitimate applicants.