AI Agents for Offensive Security: Understanding AI-powered attacks and how to stop them

Get the eBook free when you register your print book at Manning.

AI is changing how offensive security workflows are designed, executed, and analyzed. This book shows you how to build and use AI agents and multi-agent pipelines to support reconnaissance, triage, vulnerability discovery, reporting, agentic penetration testing, and red team/blue team workflows.

In this practical guide for offensive security professionals, you’ll learn how to create intelligent agents that automate parts of security testing while preserving safety, auditability, and human oversight. You’ll build your first security agent, explore how artifacts and pipelines structure agentic workflows, and see how AI can help interpret results, prioritize findings, and reduce manual triage. Along the way, AI security expert Mark Foudy emphasizes governance, authorization controls, resilience, and transparency. You’ll understand where AI adds value, and also where it introduces operational and security risk. The result is a clear framework for integrating AI into day-to-day security work.

What's inside

• Build agents for offensive security workflows
• Use LLMs for reconnaissance, triage, and vulnerability discovery
• Design multi-agent pipelines with orchestration and shared artifacts
• Improve reporting and workflows with AI assistance
• Failure modes, resilience, and auditing in agent systems

About the reader

For cybersecurity professionals, penetration testers, red teamers, detection engineers, SOC analysts, and security researchers who want to add AI to their security workflows. Familiarity with Python and offensive security concepts will help you get the most from the book.

About the author

Mark Foudy is an AI offensive security researcher, author, and neurodiversity advocate in the information security community. He specializes in adversarial AI, red teaming, and voice verification security, focusing on practical vulnerability exploitation techniques. Mark has presented research on voice verification systems at DEF CON, competed in the Social Engineering Village with an AI voice agent, and leads the DEF CON 508 chapter.