Artificial intelligence for regulatory bottleneck assessment and clustering of Brazilian gold mining processes

Brazil’s mining sector is characterized by a wide diversity of substances, stakeholders, and regulatory challenges. To promote legal traceability and responsible mineral certification, this study analyzed a national cross-sectional dataset comprising 42,705 active mining processes using an automated rule-based screening system and artificial intelligence (AI) clustering. A set of nine legal, environmental, and fiscal compliance criteria was encoded to evaluate regulatory status. The automated bottleneck analysis revealed a severe systemic deficit: only 145 processes (0.34%) fully met certification requirements. While geospatial criteria showed high compliance rates (> 93%), mandatory operational and fiscal duties such as environmental licensing, CFEM royalty payments, and Annual Mining Report (RAL) submissions emerged as the primary structural bottlenecks hindering formalization. Additionally, unsupervised AI clustering using a log-transformed K-Means algorithm (k = 3) successfully and organically segmented processes based on territorial scale and administrative maturity. The model revealed three distinct data-driven profiles: (i) large volumes of mixed/early artisanal requests, (ii) large-scale preliminary exploration titles, and (iii) a highly concentrated cluster of advanced operational permits containing all fully compliant records. The results demonstrate the analytical power of unsupervised AI and deterministic screening for diagnosing compliance gaps, supporting evidence-based regulatory enforcement, and aligning Brazil’s mineral economy with global ESG traceability standards.