01 Oct There has been a 176% Increase in Environmental Requirements in the Last Decade, Explained by More Intensive, Detailed, and Extensive Application of Existing Regulations
The study, “Use of Artificial Intelligence for Characterizing the Environmental Regulatory Burden of Projects Under Development,” published by the National Commission for Evaluation and Productivity (CNEP), utilized a generative artificial intelligence model to detail the evolution of environmental obligations over the past decade, their causes, and sectoral differences.
The results show that the number of environmental obligations faced by mining and energy investment projects in Chile increased by 176% between 2015 and 2024: while in 2015, an average project faced 90 obligations, by 2024, that figure had risen to 250.
The growth does not stem from new regulations, but rather from the more intensive application of existing rules, as well as an increase in obligations arising from voluntary commitments and ad hoc conditions defined for each project. The analysis identified 146 different regulations as sources of environmental obligations. Still, only 3.6% correspond to provisions enacted after 2015—confirming that the bulk of the increase came from historical rules applied with greater intensity.
The increase was broad-based, but proportionally higher in mining (+246%) and in small projects (+223%), which today face a level of requirements comparable to that imposed on large-scale investments ten years ago.
To process the information contained in the sample—1,336 Environmental Qualification Resolutions (RCA) issued between 2015 and 2024, documents that compile all of a project’s environmental obligations but which, due to their length and heterogeneity, had been very difficult to analyze systematically—CNEP, with the support of Stanford University professor and AI expert Gabriel Weintraub, developed a generative AI model. This innovation allowed the extraction of more than 220,000 obligations, which were organized into a public and auditable database on Chile’s environmental regulatory burden.
The study is already available on the CNEP website (www.cnep.cl). In the coming weeks, its data and results will also be accessible through an interactive dashboard to be incorporated into the institutional site.
Santiago, October 1, 2025. — As part of the seminar “Environmental Regulation in Chile: Toward Evidence to Boost Investment”, organized with the support of the Adolfo Ibáñez University Business School (UAI), the National Commission for Evaluation and Productivity (CNEP) presented the results of the study “Use of Artificial Intelligence for Characterizing the Environmental Regulatory Burden of Projects Under Development”.
The study represents an effort to systematically organize and analyze the information contained in more than 1,300 RCAs issued between 2015 and 2024. Until now, that information remained dispersed in lengthy, heterogeneous texts lacking standardization, making rigorous comparison or auditing difficult. In practice, it was a normative corpus that was complex to systematize and lacked a harmonized format, which hindered consistent analysis across time or sectors.
To address this challenge, the CNEP team, led by legal director Cristian Romero, developed and applied a Generative AI model, with the support of Gabriel Weintraub, a Stanford University expert in AI. This innovation allowed the extraction of over 220,000 environmental obligations contained in RCAs, organizing them into a unique public database on Chile’s regulatory burden. This transformed the information from a set of dispersed documents into structured, transparent, valuable evidence for public policy design.
A Decade of Growth, Stronger in Mining and Small Projects
The study reveals that mining and energy investment projects have faced a growing regulatory burden over the past decade, with the average number of obligations contained in RCAs in these sectors nearly tripling. While in 2015 an average project faced about 90 obligations, by 2024 the figure had risen to 250, a 176% increase.
The rise was not uniform across sectors:
- Mining saw the sharpest proportional increase (+246%), going from 50 to 173 obligations on average.
- Power generation rose from 104 to 268 (+158%).
- Power transmission from 92 to 237 (+158%).
Differences were also observed by project size:
- Small projects (< USD 10 million) showed the steepest increase, from 62 to 200 obligations (+223%).
- Large projects (> USD 100 million) went from 138 to 301 (+118%).
This means that today, a small-scale investment faces more requirements than those imposed only on large-scale projects a decade ago.
Where Does the Increase Come From?
The study concludes that the growth did not result from new regulations but rather from more intensive application of existing laws and from voluntary commitments and ad hoc conditions defined for specific projects.
- A total of 146 different regulations were identified as sources of environmental obligations.
- Only 3.6% correspond to provisions enacted after 2015.
- Approximately 40% of obligations originate not from laws or regulations, but from voluntary commitments by companies or project-specific conditions in RCAs.
This structural feature of the system introduces flexibility but also unpredictability, limiting regulatory certainty for investment planning.
What Kind of Obligations Predominate?
- 84% are “means obligations” (specifying how an objective must be met).
- 16% are “results obligations” (stating only the goal).
- This shows Chilean environmental regulation focuses more on process than outcome, which facilitates oversight but increases rigidity and administrative burden.
Among the 50 Sectoral Environmental Permits (PAS) defined by law, more than 60% are required in the cases analyzed—mainly those related to waste management, subdivision of rural land, and protection of archaeological and paleontological heritage.
Methodological Innovation and AI Projections
Generative AI made it possible not only to process RCAs but also to identify, classify, and standardize each environmental obligation, transforming heterogeneous texts into a coherent, comparable inventory.
For the first time, this allows precise analysis of the origin, magnitude, and evolution of Chile’s environmental regulatory burden. Results are organized into a public, auditable database, ensuring that researchers, institutions, and citizens can verify and reuse the data with confidence.
An interactive dashboard (to be launched at www.cnep.cl) will enable users to explore data, compare sectors and projects, and visualize trends over time. This marks a milestone in regulatory transparency and access to information.
CNEP now considers extending this methodology to other sectors with large volumes of unsystematized regulatory information. Applying the generative AI model elsewhere could help organize and audit regulatory evidence, strengthening regulatory quality and supporting public policy decisions based on empirical evidence.