AI-Driven Fraud-Risk Analytics and Risk-Responsive Auditing in U.S. Public Companies Financial Reporting: A Structured Integrative Review and Signal-to-Assurance Framework

Artificial intelligence (AI) and data analytics can expand auditors' ability to identify unusual transactions and reporting patterns, but their value depends on what happens after a risk signal is generated.This structured integrative review examined how AI-generated fraud and material-misstatement risk signals translate into auditor judgment, riskresponsive procedures, and assurance outcomes in U.S. public company auditing.Literature identification covered peerreviewed, English-language articles published from January 2010 through August 18, 2026, and combined scholarly and publisher searches, prior reviews, and backward and forward citation chaining.Studies were eligible if they addressed signal validity, auditor interpretation or reliance, conversion of recognized risk into audit procedures, or assurance outcomes.Work confined to internal audit, nonfinancial fraud, or technology unconnected to external financialstatement auditing was excluded, as were preprints and vendor materials.Twenty core empirical studies were retained for the primary synthesis and extracted into a structured evidence matrix; Public Company Accounting Oversight Board standards formed a separate regulatory corpus.Seven core studies addressed signal validity, six auditor interpretations or reliance, six planned or performed audit responses, and one assurance-related outcome.Evidence was concentrated on risk identification and auditor response, with much less evidence on procedural change and assurance outcomes.The only core study examining an assurance-related outcome assessed firm-level AI investment rather than a specific analytical signal.The Four-Gate Signal-to-Assurance Framework was developed to show where those breaks in the evidence occur.The literature does not yet show what happens when a specific AI-generated risk signal is followed through to an audit-quality outcome.