Integrating Generative AI Into HSE Legal Compliance: Contributions, Limits and Organisational Impacts

This article analyzes the contribution of generative artificial intelligence (GAI), and more specifically large language models (LLMs), to legal work in the field of health, safety, and environmental (HSE) compliance within a mediumsized LegalTech company. The study focuses on a French SME specializing in regulatory monitoring, compliance management, and HSE risk assessment.
The research is based on an action research approach, combining field observations, interviews with HSE legal experts, and experiments using a proprietary database of legal questions. Three complementary evaluations were conducted: a qualitative evaluation of the generated responses, an evaluation of the impact of prompt engineering, and a quantitative performance analysis.
The results demonstrate that GAI significantly improves the responsiveness and efficiency of regulatory monitoring and information retrieval, with performance varying depending on the legal area. The structured design of prompts considerably improves the consistency, accuracy, and usefulness of the generated responses. However, limitations remain regarding reliability, risk of error, interpretability, and the need for expert validation. Integrating GAI also raises strategic, organizational, and ethical challenges within legal teams.
The study concludes that a gradual and structured integration of GAI, supported by a new organization of legal work, can improve the performance of an HSE legal department while creating sustainable organizational value, provided that human expertise remains central.

Keywords: AI, LLM, GAI, Compliance.