AI-Assisted Exploit Development

Using Large Language Models to Accelerate Vulnerability Research, Reverse Engineering, and Offensive Security

Dedication

Introduction: The New Exploit Developer’s Toolkit

1. How This Book Is Organized
2. Who This Book Is For
3. What This Book Does Not Cover
4. A Note on Ethics and Responsibility
5. How to Use This Book

Chapter 1: The New Exploit Developer’s Toolkit

1. The New Reality: What Has Changed
2. A Second Case: Finding a Linux Kernel Zero-Day with o3
3. A Third Case: Autonomous Exploit Generation Against Modern Mitigations
4. What AI Does Well: The Force Multiplier Effect
5. Thinking About AI as a Tool, Not a Replacement
6. A Note on Ethics and Responsibility

Chapter 2: Foundations of AI-Assisted Security Research

1. How Transformers Work: Enough to Understand What They Can and Cannot Do
2. Prompt-Based Inference vs. Fine-Tuning: Choosing Your Approach
3. Prompt Engineering Paradigms for Security
4. The Trust Boundary: Formalizing Human-AI Interaction
5. Key Principles for Working with AI in Security Workflows
6. Hands-On Lab: Evaluating LLM Output Reliability

Chapter 3: The Evolution of Automation in Security Research

1. The Early Era: Signature-Based Detection and Rule Matching
2. The Structural Analysis Era: Static and Dynamic Program Analysis
3. The Generative AI Era: LLMs as Security Co-Pilots
4. The Quantitative Leap: What AI Adds to Each Era
5. Why AI Represents a Qualitative Shift
6. The Limits of Automation: What No Paradigm Can Do
7. Hands-On Lab: Tracing the Evolution of Automation

Chapter 4: Setting Up an AI-Augmented Research Environment

1. The Landscape of AI Tools for Security Research
2. Local LLM Deployment Options
3. Cloud API Considerations
4. Integrating AI with Reverse Engineering Platforms
5. The Model Context Protocol: Connecting AI to Tools
6. Operational Security for AI-Assisted Workflows
7. Hands-On Lab: Deploying a Local AI Research Environment

Chapter 5: AI in Binary Analysis and Reverse Engineering

1. Automated Function Identification and Naming
2. Decompilation Refinement with LLMs
3. Guided Reverse Engineering: Asking the Right Questions
4. Case Study: AI-Assisted Analysis of XLoader 8.0
5. Limitations and Failure Modes
6. Hands-On Lab: AI-Assisted Binary Analysis with Evidence-First Prompting

Chapter 6: Vulnerability Discovery with AI and Fuzzing

1. The State of AI-Guided Fuzzing
2. Key Research: FuzzGPT, CHATAFL, TitanFuzz, and Beyond
3. Performance Comparison: Traditional vs. AI-Guided Fuzzing
4. The Validation Problem
5. Static Analysis Augmentation with AI
6. Web Application Vulnerability Discovery
7. Case Study: Sean Heelan’s CVE-2025-37899 Discovery
8. Hands-On Lab: AI-Guided Protocol Fuzzing with CHATAFL Principles

Chapter 7: Proof-of-Concept Exploit Development

1. The Landscape: AI-Powered Exploit Development Platforms
2. Shellcode Generation
3. ROP Chain Construction
4. Mitigation Bypass Techniques
5. Case Study: Sean Heelan’s QuickJS Exploit Experiments
6. Safety, Ethics, and Responsible Disclosure
7. Hands-On Lab: Building an AI-Assisted Shellcode Test Harness

Chapter 8: Malware Analysis and AI-Assisted Triage

1. The Malware Analysis Pipeline
2. AI-Powered Initial Triage
3. Benchmarking LLMs for Malware Triage
4. Automated YARA Rule Generation
5. Unpacking and Deobfuscation with AI
6. Case Study: Rapid Triage of a New Malware Family
7. Limitations and Risks
8. Hands-On Lab: AI-Assisted Malware Triage with YARA Generation

Chapter 9: Web Application Exploit Development with AI

1. The Web Application Attack Surface
2. Burp Suite and AI Integration
3. AI-Assisted Vulnerability Discovery Workflows
4. Common Pitfalls and How to Avoid Them
5. Hands-On Lab: AI-Assisted Web Application Vulnerability Discovery

Chapter 10: Advanced Workflows and Automation

1. Building AI-Powered RE/ED Pipelines
2. MCP-Enabled Tooling: Connecting LLMs to Debuggers and Disassemblers
3. Automated Report Generation and Documentation
4. Multi-Agent Workflows: Orchestrating Multiple AI Models
5. Case Study: End-to-End AI-Assisted Vulnerability Research Pipeline
6. Scaling Your Workflow: From Individual Researcher to Team Operations

Chapter 11: Operational Security and Risk Management

1. The Data Flow Problem: What Goes Where, When, and Why
2. Practitioner Pushback: Why Many Reverse Engineers Refuse AI
3. Operational Security Protocols: A Practical Framework
4. The Economic Reality of AI-Assisted Security Work
5. Legal and Ethical Considerations
6. Hands-On Lab: Building an OpSec-Secure AI Research Environment

Chapter 12: The Future of AI-Assisted Exploit Development

1. Multimodal Models and Vision-Based Reverse Engineering
2. Specialized Security LLMs: The Domain-Specific Revolution
3. The Arms Race: AI-Assisted Attacks vs. AI-Assisted Defense
4. Skills That Will Remain Valuable as AI Matures
5. What the Next Five Years Look Like
6. Preparing for the Future: A Strategic Framework

Chapter 13: Becoming the AI-Augmented Expert

1. The Expert’s Mental Model: A Decision Framework
2. The Iterative Refinement Loop
3. Building Your Personal AI Toolkit
4. The Professional’s Code: Ethical Guidelines for AI-Assisted Security Research
5. The Long Game: Sustaining Effectiveness Over Time

Chapter 14: Conclusion

References

Appendix A: Quick Reference Card — Model Selection Guide

Appendix B: Prompt Templates — Quick Reference

Appendix C: Common Failure Modes — Quick Reference

Appendix D: Glossary of Key Terms