Human-in-the-Loop Generative AI Frameworks for Reliable Financial Reporting in Enterprise Systems
Ramsha Siddiqui · Zenodo (CERN European Organization for Nuclear Research) · 2026
This study introduces a Generative AI Human-in-the-Loop (HITL) Framework that facilitates the reliability, transparency, and compliance of financial reporting systems at enterprises. This HITL framework facilitates financial statement preparation, revenue summarization, auditing, and disclosures through the assistance of AI while ensuring structural human control. The proposed framework comprises elements such as confidence-based intervention, expertise-aware validation, explainable AI, auditing traceability, and strategic escalation that help rectify financial reporting mistakes, issues associated with uncertainties in AI technology, and compliance problems. Furthermore, the framework consists of narrative data validation to ensure consistency between data and reports generated.