人工智能大模型在财会监督中的应用场景与风险规制研究

Research on Application Scenarios and Risk Regulation of Large-Scale Artificial Intelligence Models in Financial and Accounting Supervision

摘要: 随着人工智能技术突破，大模型正深刻变革财会监督领域。本文采用系统性文献综述方法，研究发现，现有文献表明大模型在自动化审计、欺诈检测、文本分析与智能决策等场景中展现出提升数据处理效率和辅助穿透式监管的潜力；然而，技术应用也伴随数据隐私泄露、算法偏见固化、责任归属困境等风险。初步证据显示，当前大模型在财会监督中的实际应用仍处于探索阶段，其能力边界、可靠性与可解释性尚需进一步验证。结论认为，需通过完善数据治理、加强算法审计、明确伦理边界并推动监管科技同步发展，促进技术与监督的良性融合。

Abstract: With breakthroughs in artificial intelligence technology, large language models are profoundly transforming the field of financial and accounting supervision. This paper employs a systematic literature review methodology and finds that existing research indicates large language models demonstrate the potential to enhance data processing efficiency and support penetrative supervision in scenarios such as automated auditing, fraud detection, text analysis, and intelligent decision-making; however, the application of this technology also carries risks such as data privacy breaches, the entrenchment of algorithmic bias, and the dilemma of assigning responsibility. Preliminary evidence suggests that the practical application of large language models in financial and accounting supervision remains in an exploratory phase, and their capabilities, reliability, and interpretability require further validation. The conclusion argues that a positive integration of technology and governance must be fostered by improving data governance, strengthening algorithmic auditing, clarifying ethical boundaries, and promoting the synchronized development of regulatory technology.