TL;DR:

A practical discussion with security leaders on what frontier AI models like Mythos can actually do today, where they create real security value (and risk), how they may reshape security programs, budgets, and talent strategies, and what it means when defenders and attackers have access to the same capabilities.

Frontier AI models like Claude Mythos are no longer a research curiosity — they're writing code, finding vulnerabilities, and reshaping how security work gets done. But what do they practically mean for the people responsible for defending organizations?

Join Vinnie Liu (Bishop Fox), Jason Lish (Global CISO, Cisco), and Adrian Peters (Portfolio CISO, Vista Equity Partners) for a candid, experience-driven conversation — not a vendor pitch. Drawing on hands-on work with Mythos and other frontier and open-weight models, the panel will cover what these models do well today, where they fall short, how they fit (or don't) into your existing security stack and budget, what they mean for your talent strategy, and how to think about the same capabilities landing in attackers' hands.

Whether you set security strategy or live in the SDLC every day, you'll leave with a grounded view of where this technology is, where it's headed, and how to start preparing your organization now.

Questions we'll cover:

1. Is Mythos a genuine inflection point for security, or just the latest frontier model — and how does it stack up against open-weight alternatives on bang for your buck?
2. What have these models actually done in our hands — what's real, what's hype, and what's improving fastest?
3. Where does this fit in your SDLC and security stack — and given token costs, which existing controls start to lose relevance?
4. If budget forced a tradeoff tomorrow, what would you cut first: SAST/DAST, manual AppSec, pentesting?
5. Does this push security ownership further into engineering?
6. Can your average developer or analyst get real value from these tools, or does it take specialized talent to unlock them?
7. How does an organization actually start adopting this — and what are the roadblocks that stall it?
8. What happens when attackers are using the same models?