AUDITFLOW: Executable Symbolic Environments for Structured Financial Reporting Verification
Researchers have developed AuditFlow, a novel multi-agent framework designed to enhance the verification of structured financial reports using AI. This system creates a symbolic environment by integrating US-GAAP taxonomy and XBRL filing data, enabling agents to perform complex tasks like fact retrieval, taxonomy traversal, and numerical checking. AuditFlow achieved 82.09% joint audit accuracy on a financial auditing dataset, significantly outperforming existing baselines and demonstrating the critical role of its deterministic verification component. AI
IMPACT Enhances AI capabilities in structured financial auditing, potentially improving accuracy and efficiency in regulatory compliance.