2026-10-31 –, Track 4 (207) Language: English
Claude Mythos is finding zero-days in production software. Project Glasswing is letting AWS, Apple, Google, Microsoft and JPMorgan use it to pre-patch critical vulnerabilities before attackers get there. The rest of us are stuck with AI tools that refuse to analyze a suspicious PowerShell command.
That gap is real, it's frustrating, and it's fixable.
Most security practitioners have formed their opinion about AI from one of three places: a vendor pitch, a ChatGPT refusal, or a breathless blog post. None of those tell you what AI can actually do for your work today. This workshop does.
Vibe hacking is the practice of directing AI through security tasks using your expertise and judgment as the guide. The model handles the volume. You handle the judgment. It is not about replacing your skills. It is about multiplying them.
Over the course of the day, you will use a purpose-built security AI platform to analyze logs, prioritize vulnerabilities, review code for exploitable flaws, assist with web application penetration testing, and develop proof of concept exploits. You will understand how these tools work under the hood, why commercial AI refuses to help with legitimate security tasks, and how the techniques you practice today apply to any capable agentic AI platform, not just the one we use in the workshop.
No coding required. No passive demos. You do the work, you see the results, and you leave with a clear-eyed answer to the question the whole industry is arguing about right now: what can AI actually do for security practitioners today?
This workshop addresses a real gap: security practitioners who hear about AI constantly but have never used a capable tool to do actual security work. Commercial AI products have safety guardrails fine-tuned into their weights, not just their system prompts. They refuse to analyze malware, assist with exploit development, or help with adversarial code review, even for fully authorized practitioners. This is not a prompting problem. It is a weight-level behavior that cannot be reliably overridden.
The workshop uses Hacker Sidekick, a purpose-built security AI platform with integrated tooling (Shodan, DNS, CVE databases, PCAP analysis), to demonstrate what a capable, unconstrained tool can do. The techniques and workflows taught apply to any agentic AI security platform. This is not a product pitch.
Presenter: Robert Wagner, Advisory CISO, Security Strategist and Architect, co-organizer of BurbSec, co-founder of Hak4Kidz. Deep practitioner background across red team, blue team, GRC, and executive advisory work.
Number of instructors: 1 lead, with optional assistants for high-enrollment sessions.
Outline/Notes:
Morning: AI Foundations
Welcome and Introductions
What Is Vibe Hacking and Why Do It?
Security work is cognitive work - where the bottleneck actually is
The model handles volume. You handle judgment.
How LLMs Actually Work
Tokens, vectorization, attention, context windows
Training vs. inference vs. fine-tuning and why it matters for security
Hallucination: not a bug, a structural property
The Guardrails Problem
Why commercial AI refuses legitimate security tasks
Prompted vs. fine-tuned behavior: why you cannot prompt your way around it
Live comparison: commercial AI vs. purpose-built alternative
What It Would Take to Build This Yourself
Component list: model access, legal/policy, tool integrations, orchestration, infrastructure
DIY homelab edition: open source path with real trade-offs
Prompting: Foundations and Advanced Techniques
Role, context, output shape, uncertainty rules
Chain-of-thought, few-shot examples, iteration strategies
Weak vs. strong prompt live comparison
Workshop Setup Verification
Afternoon: Hands-On Scenarios
Scenario 1: Log Analysis Scenario 2: Vulnerability Prioritization Scenario 3: Static Code Analysis (includes adversarial framing commercial AI refuses) Scenario 4: Web App Penetration Testing Support Scenario 5: Exploit Development - CVE and research-based proof of concept exploits
Supplemental (self-guided or time permitting) Scenario 6: Threat Actor Profiling Scenario 7: SOC Alert Investigation - bring your own data
Wrap-Up and Q&A
The Foundations section is about 2 hours, and the hands on exercises are another 2 hours.
Robert Wagner is an Advisory CISO and highly respected security
practitioner, advisor and strategist with over 20 years of experience.
His security experience ranges from defending everything from Fortune
500 companies to government agencies, major universities, and
financial institutions. He is a co-founder of the not-for-profit
organization Hak4Kidz, serves on the board of the Chicago ISSA
chapter, and regularly volunteers for various hacker cons including
Bsides312, BlueTeamCon, and others.