BSides Knoxville 2026

It's our 12th year! We'll chat a bit about:

• Organizer changes (and availabilities)
• How our finances are looking headed into 2027
• Introduce our platinum sponsors (Clayton Homes, Cisco ASIG, and Starseer)

Your office printer is a weapon. Watch how attackers weaponize forgotten network devices to bypass MFA, steal tokens, and infiltrate enterprise environments. Discover the attack chain, why defenses fail, and how to stop it.

Who is this talk for?
Security folks, tool builders, and tinkerers who are using (or thinking about using) AI in their workflows—and have already noticed it can be equal parts genius and nonsense.


Problem Statement
LLMs are being dropped into security tooling everywhere—writing scripts, reviewing findings, generating reports, even helping with exploitation. The problem is, they don’t actually know anything. They predict what sounds right.

And sometimes what sounds right… ain’t right.

The real danger isn’t obvious failure. It’s when the output looks clean, reads well, and passes a quick glance—but is fundamentally wrong in a way that can waste time, introduce risk, or quietly break your workflow.

This talk takes a practical, slightly hillbilly approach: trust your tools, but verify everything they do.


Key Takeaways
AI will be wrong; and it will be confident about it
“Looks right” is one of the most dangerous failure modes in security
You need deterministic validation, not vibes
Treat AI like a junior: useful, fast, but needs supervision
A little hillbilly common sense; “that don’t smell right” is still one of your best defenses

For nearly a decade, I've been gathering and studying details from breaches. In the beginning, there wasn't much, but over the years, more and more has emerged in the public record. There's enough data now that it's time to start talking about what we can learn from the failures of others.

This will follow up on my RSAC 2026 talk, which focused more on making the argument for breach transparency. This talk will focus almost exclusively on the specifics I've learned from a decade of studying breaches, which the RSAC talk only shared a sample of.

Check out Hamburger Hill or make a suggestion for next year's food truck!

Described since 2011, the security poverty line is that point below which an organization can’t effectively secure itself. This is due to a lot of factors, and the recent popular trend of AI use (particularly LLMs) has muddied the waters even more. Is it going to save the vast majority of security have-nots? Will it only help the token-rich 1%? Or is it mostly just helping the attackers?

In this potentially contentious session, we’ll talk about what’s really within reach.

Every organization has a human attack surface. Most have no idea how big it is. Using basic scraping methods against public social media platforms, we built a graph database that disambiguates profiles across platforms and resolves fragmented online identities into confirmed employees. The graph maps organizational hierarchies, team structures, reporting relationships, and personal details. This talk covers the architecture, the data sources, and how the disambiguation engine works at scale, then shows what an entire organization looks like when you query it through the lens of someone building a target list.

Armed with intelligence from the graph, we deployed an AI voice agent to call a Fortune 100 help desk. The agent passed identity verification, convinced the analyst to initiate a password reset, and achieved account takeover without any technical exploitation. We'll play the distorted call recording and break down exactly where the verification process fell apart.

Help desks were designed for a world where the caller was human and the information they had was hard to get. Neither of those things is true anymore. We'll cover practical steps security teams can take to harden their help desks, reduce their human attack surface, and prepare for a threat model where the attacker already knows everything about your employees before they ever pick up the phone.

How cybercrime speaks matters. This talk compares English, Spanish, and Russian underground forums, revealing how language shapes access, hierarchy, recruitment, targets, and tactics, from open global hubs to regional networks and tightly controlled elite communities.

Traditional offensive security is often a battle against the friction of manual reconnaissance, repetitive infrastructure setup, and the tedious translation of tool outputs into actionable reports. As Large Language Models evolve from static chat interfaces into autonomous agents, the role of the security professional is shifting from Manual Operator to Systems Architect.

This session approaches offensive security through the lens of a deckbuilding card game. We will walk through a "Setup" phase for selecting the right agentic infrastructure, an "Action" phase involving C2 infrastructure and application security use cases, and a "Cleanup" phase using practical, human-in-the-loop workflows to finalize deliverables.

- Combos and setup: Automatically provisioning multi-cloud redirectors across Cloudflare, AWS, and Azure while maintaining origin secrecy.
- Know your tools: Chaining playwright-mcp with automated script generation to transform a browser-based discovery into a persistent exploit.
- Manage your mana: Strategies for maintaining a low-cost operation by "compiling" agentic reasoning into local CLI tools and offloading repetitive tasks to local models like Ollama.

Attendees will leave with a blueprint for building their own "Security Deck," enabling them to automate the grind, preserve high-level strategy, and execute complex engagements at machine speed without losing the critical oversight of the human operator.