Table of Contents
As a cybersecurity operator, I’ve found Large Language Models (LLMs) to be a small revelation in terms of testing initial hypotheses, identifying blind spots and knowledge gaps, and accelerating research into new technologies. Used correctly, they’re great sparring partners.
To help you unlock similar value, here are a set of questions I recommend anyone in the cybersecurity space pose to their favorite AI chatbot. These prompts are designed to spark interesting, practical conversations and ultimately lead to new avenues for security research and operational improvement.
To get the most value from these prompts, we recommend adapting them with prompt engineering best practices. Feel free to prime your LLM AI chatbot with specific details about your unique cybersecurity deployment, ask the chatbot to assume the persona of an expert cybersecurity advisor, or specify the format of the output you would like to see.
1. Visibility / Continuous Compromise
“Outline the non-obvious steps a successful attacker takes after breaching the perimeter but before their presence is officially detected by security tools.”
2. Breach Detection Gap / Dwell Time
“Create a list of 5 key indicators that a persistent threat actor has been operating inside a corporate network for over 60 days without being flagged by typical EDR or SIEM tools.“
3. SecOps Efficiency / Alert Fatigue
“If a security analyst receives 1,000 alerts a day, describe a system that could reliably reduce that number to the 5 most conclusive and actionable events without losing critical data.”
4. Attack Surface Assessment / Shadow IT
5. EDR Evasion / Network-Level Compromise
“Explain the concept of ‘living off the land’ attacks and how they are designed to deliberately bypass endpoint detection mechanisms by using built-in system tools.”
6. Threat Intelligence / Context
“Detail the ideal structure for a real-time threat intelligence feed that is immediately usable by both automated response systems and human analysts, focusing on context over volume.”
7. Incident Management / Prioritization
“Develop a five-level priority ranking system for security incidents based not just on the type of malware, but on the proven contact and exposure level of the affected assets.”
8. Integrated Response / Maximizing Stack
“Describe an ideal automated workflow where a firewall, an EDR, and a cloud access security broker (CASB) share confirmed compromise data in real-time to simultaneously block a threat.”
9. Retrospective Compromise
“A new, severe zero-day is announced. What data would a CISO need to confidently confirm their environment has never been exposed to this threat’s indicators of compromise (IoCs) over the past two years?”
10. Measuring Effectiveness
“Beyond annual penetration tests, propose three continuous, objective metrics an organization could use daily to truly measure its overall level of compromise rather than just its compliance.”
