# AI Safety & Guardrails Engineering for Government Systems (Advanced) (Self-Paced)

Canonical URL: <https://www.graduateschool.edu/courses/ai-safety-guardrails-engineering-for-government-systems-advanced-self-paced>

## Overview

Building AI for government means engineering the controls around it. This advanced, self-paced course covers the practical safety engineering of AI systems - input and output filtering, prompt-injection defense, red-teaming, evaluation harnesses, and monitoring - so the systems you build behave safely and predictably in production.

## What you'll learn

- Identify safety risks in AI systems.
- Engineer input and output guardrails.
- Defend against prompt injection and misuse.
- Red-team an AI system.
- Build evaluation and monitoring.
- Plan a safety approach for a government AI system.

## Prerequisites

Programming and AI-building experience expected.

## Curriculum

#### Module 1: What can go wrong

- AI failure and abuse modes
- Prompt injection and jailbreaks
- Government-specific risks
- A safety mindset

#### Module 2: Building guardrails

- Input validation and filtering
- Output checks and constraints
- Limiting tool and data access
- Defense in depth

#### Module 3: Red-teaming and evaluation

- Red-teaming your own system
- Building an evaluation harness
- Measuring safety and quality
- Catching regressions

#### Module 4: Monitoring and response (capstone)

- Monitoring in production
- Incident response for AI
- Documenting safety
- A safety plan for your system

## Pricing

**Tuition:** $899