SEC536: Adversarial AI - Penetration Testing AI Systems

Virtual
A SOC's outcomes mirror the program around it. When asset context, identity data, exposure and business priority live with other teams, analysts work without the full picture and every handoff slows them down. So how can the SOC work smarter and faster?
The answer is automation that extends across the whole security program, both upstream and downstream of the SOC. In this session, we’ll examine where agents and workflows can take on operator roles and keep context flowing between teams. The result is a SOC that makes better decisions and moves faster, because it has the context of the broader program at the speed of AI.
Attendees will leave with a way to map their program's dependencies, a framework for deciding what to automate and how, including when to use agents and when to use determinism, and a realistic view of what agents can do today and what still needs a human.
*Sponsored by BlinkOps
Virtual
While many native security operations agents dramatically improve security workflows, organizations sometimes need more customization.
This session will demonstrate how to build and orchestrate custom SOC agents using Google Security Operations MCP servers and the Agent Development Kit (ADK). You will learn how to structure agent instructions, enforce security guardrails, and reason over security logs, incidents, and threat intelligence.
*Sponsored by Google
Virtual
Every vendor now claims an "AI SOC," and AI-accelerated threats keep raising the stakes. Yet most solutions stop at triage, which is where the real work begins. A true AI SOC runs the complete threat lifecycle end to end, and the real differentiator is the layer beneath the agents: the context that grounds every decision, the memory that gives it precedent, and the learning that teaches it your team's judgment.
See how that comes together across the Torq AI SOC Platform, where Torq Auto Triage cuts EDR noise up to 97% and works with Case Management, Socrates, and Hyperautomation to slash MTTR 94% on average. The Context Graph and Recall give the SOC lasting memory, while Reflex learns your team's judgment to become uniquely yours, with a 20-point accuracy lift over the nearest alternative.
Join Torq's Rick Bosworth as we discuss how a true AI SOC works and what security leaders should consider before betting on one. Walk away with: What defines a true end-to-end AI SOC, and why context, not knowledge, is the line that matters How context, memory, and learning combine into an AI SOC that compounds in value the longer it runs A CISO's playbook for the shift to agentic SecOps: where to start with autonomy, how the org chart changes, and how to build trust as you scale
*Sponsored by Torq
Virtual
Virtual
Every vendor now claims an "AI SOC," and AI-accelerated threats keep raising the stakes. Yet most solutions stop at triage, which is where the real work begins. A true AI SOC runs the complete threat lifecycle end to end, and the real differentiator is the layer beneath the agents: the context that grounds every decision, the memory that gives it precedent, and the learning that teaches it your team's judgment.
See how that comes together across the Torq AI SOC Platform, where Torq Auto Triage cuts EDR noise up to 97% and works with Case Management, Socrates, and Hyperautomation to slash MTTR 94% on average. The Context Graph and Recall give the SOC lasting memory, while Reflex learns your team's judgment to become uniquely yours, with a 20-point accuracy lift over the nearest alternative.
Join Torq's Rick Bosworth as we discuss how a true AI SOC works and what security leaders should consider before betting on one. Walk away with:
*Sponsored by Palo Alto Networks
Virtual
AI is transforming the SOC by helping security teams analyze threats faster, automate investigations, and reduce the burden of repetitive tasks. This session explores how AI can augment analysts, accelerate malware and phishing analysis, and enable more efficient, scalable threat detection and response.
*Sponsored by VMRay
Virtual
A defender can still outthink an attacker, but they can't out-work one that runs on software. Closing that gap means using AI to automate defense, but before teams can trust AI with their most critical security problems, they need to know it will work when they need it to. AI resilience means having models that are still available, even if a provider, service, or API isn't.
This session introduces AlertZero, Elastic Security’s agentic layer, and the three key controls teams need for AI resilience:
When used effectively, AI doesn't replace analysts, it accelerates their triage, investigation, and threat hunting. The human isn’t removed from the loop, they’re moved to the top of it, making the final call on the SOC's most important decisions.
Learn the questions to ask any AI security vendor, including Elastic, before trusting agents with critical security operations decisions.
*Sponsored by Elastic Security
Virtual
Virtual
The CIS Security Operations Center (SOC) is the unseen key to its Services. In this presentation, we'll look at how the SOC is the foundation of what we deliver and threads cross-agency intelligence together to detect threats sooner, respond faster, and protect organizations of every size and sector.
We'll walk through how the SOC powers and draws on CIS Managed Detection and Response (MDR), Managed Extended Detection and Response (MXDR), Cyber Threat Intelligence, the Cyber Incident Response Team (CIRT), and Red Team services.
You'll see how detection, intelligence, and response come together, and how what we learn in one area makes every other service stronger. You'll leave with a clearer picture of how these services fit together, what happens behind the scenes when an alert fires, and how to get the most from your CIS relationship.
*Sponsored by Center for Internet Security
Virtual
AI didn't invent cyberattacks, but it has significantly reduced the cost, skill, and time they used to take. The gap between a patch shipping and a working exploit has shrunk from weeks to under an hour. Attackers are also running model-directed malware like LameHug, which generates fresh commands on the victim's machine every run. No two runs look alike, so signatures don't catch it.
You can't prevent your way out of this. You have to be able to see it. To survive the post-Mythos world, organizations need an Agentic SOC. For the practitioner doing the work every day, here's what that means in practice.
This session is a 101-level playbook for starting your Agentic SOC journey. We'll cover the pitfalls we see teams hit most often and the quick wins you can take back to your team right away. We'll also make the case that network visibility should be your ground truth: watching the wire shows you every conversation without installing an agent. Then we'll drop the slides for a live demo of token-efficient reasoning and behavioral detection hunting threats at machine speed.
*Sponsored by Extrahop
Virtual
*Sponsored by Fortinet
Virtual
Virtual
*Sponsored by Sentinel One
Virtual
*Sponsored by BlinkOps and Google
Virtual
Virtual