AI-Enhanced Facility Finder

Overview

Development stage

PilotSource: b) Pilot – The use case has been deployed in a limited test or pilot capacity.

High-impact designation

Not high-impactSource: c) Not high-impact

Impact justification

The output of this AI use case does not serve as a principal basis for decisions or actions that have a legal, material, binding, or significant effect on rights or safety.

Start date

2025-07-01 (day precision)Source: 07/01/2025 00:00:00

Withheld from public reporting

NoSource: a) No

Mission

Topic area

Law Enforcement

Operational functions

Risk scoring and triage Derived by keyword rules; see methodology

Problem the AI is intended to solve

Many facilities are required to register for industrial stormwater permits but do not do so. To find unpermitted facilities, EPA personnel use approaches that are time-consuming, inefficient, and infeasible to scale to the millions of facilities in the U.S.

Expected benefits

AI-Enhanced Facility Finder saves time for EPA employees by identifying facilities most likely to lack necessary industrial stormwater permits, which are then reviewed by humans. Testing at the state level has shown these identifications increase the number of facilities that are permitted and monitored, resulting in cleaner water for Americans.

System outputs

The model outputs the geolocation of facilities at increased likelihood of needing an industrial stormwater permit but lacking one. It also outputs a risk score, from “No risk” to “High Risk” for permitted facilities, which is used as one of many inputs by human reviewers.

Technology

AI classification

Computer visionSource: Computer Vision: AI that processes and interprets visual data (e.g., images and videos).

System name(s)

Not reported

Custom-developed code

Yes

Public source code

Not reported

Sourcing

How it was built

Developed in-houseSource: b) Developed in-house

Vendor (as reported)

Not reported

Vendors (standardized)

None on the standard list

A vendor is the supplier named by the agency. It does not identify the underlying model or AI technology.

Data and privacy

Involves PII

No

Privacy Impact Assessment

Not reported

Authorization to Operate

No

Demographic features

a) Race/Ethnicity; c) Age; e) Socioeconomic Status; i) Income

Training and evaluation data

Data from the following datasets were utilized in conjunction with EPA permitting data and the EPA-derived Address Comparison Tool (ACT) to score the likelihood of a facility not having an industrial stormwater permit. The model was trained by utilizing previous targeting efforts by technical staff before the development of the model and was tested with various trial runs before releasing as a pilot to test its effectiveness in practice. - EPA and state permitting data (e.g., address, permit status). - Facility information via paid subscription with Dunn & Bradstreet (e.g., number of employees, facility size, revenue, corporate hierarchy, industrial sector, and credit risk indicators). - Employment data from the Longitudinal Employer-Household Dynamics (LODES). - Socioeconomic indicators from the American Community Survey data (ACS, 2017-2021). - Distance to waters by using the National Wetlands Inventory. - Elevation from the USGS Elevation Point Query Service (EPQS) - Aerial satellite imagery

Federal Data Catalog

Not reported

Governance

0 of 8 minimum-practice questions answered. A blank answer means the agency reported nothing; it does not mean the practice is absent.

Potential impacts and how they were identified

Not reported