Mythos Special: A Big Bug Problem with Gadi Evron, Rob Lee and Ed Skoudis | 29

In this episode, Ciaran and James sit down with Gadi Evron, Rob Lee and Ed Skoudis to discuss Claude Mythos and the paper co-authored with SANS, the Cloud Security Alliance, [un]prompted and OWASP in response. Gadi, Rob and Ed share their expertise on AI-driven vulnerability discovery, how cyber teams should respond, and what this shift means for the future of security work.

Gadi Evron is the founder and CEO of Knostic and CISO in residence for the Cloud Security Alliance. Rob Lee is Chief AI Officer and Chief of Research at SANS. Ed Skoudis is the founder of the SANS Penetration Testing Curriculum and Counter Hack and President at SANS Technology Institute.

Highlights:

The Mythos announcement and why it matters

What Anthropic announced with Mythos and why it’s raising concerns.

• Assessing Claude Mythos Preview's cybersecurity capabilities
• Six Reasons Claude Mythos Is an Inflection Point for AI—and Global Security | Council on Foreign Relations
• Project Glasswing: Securing critical software for the AI era \ Anthropic
• Our evaluation of Claude Mythos Preview’s cyber capabilities | AISI Work

Industry response and what’s changing

How the cyber community is reacting, what people are already seeing in practice, and how AI is starting to change the way the work itself gets done.

• SANS Institute, Cloud Security Alliance, [un]prompted, and OWASP | CSA | The “AI Vulnerability Storm”: Building a “Mythos-ready” Security Program Paper
• The AI Vulnerability Cataclysm is Coming: What It Is and How We Prepare
• Joshua Wright - RSAC2026 SANS Panel - AI-Generated Zero Day Exploits and Supply Chain:Your Vendor's Vendor's Vendor
• Dan Guido - 200 Bugs/Week/Engineer: How We Rebuilt Trail of Bits Around AI | [un]prompted 2026
• Short term pessimistic. Wildly long term optimistic. | Phil Venables
• SANS Critical Advisory: BugBusters - AI Vulnerability Discovery Hype versus Reality AI Vulnerability Discovery Is Already Here. We Have Been Doing It for 15 Months. | SANS Institute
• How the Vulnerability Operations Centre serves as an essential mission control for vulnerabilities | Digitalisation World

Tools and agents mentioned

Some of the tools discussed as a way to start getting hands-on with this now.

• OpenAnt: Find Real Vulnerabilities with LLM-Powered Analysis | Knostic
• Cursor
• Microsoft Copilot
• Introducing Aardvark: OpenAI’s agentic security researcher
• Codex | AI Coding Partner from OpenAI
• Claude Code by Anthropic | AI Coding Agent, Terminal, IDE

Additional Resources:

• Trusted access for the next era of cyber defense | OpenAI
• Easterly: Cybersecurity is a software quality problem | CyberScoop
• The AI Doc
• Disrupting the first reported AI-orchestrated cyber espionage campaign | OpenBSD TCP SACK vulnerability patch

Subscribe & Follow:

• Subscribe to the podcast on Apple Podcasts, Spotify, or your favorite podcast platform.
• Follow us on X, Facebook, and LinkedIn for updates.

Contact:

• Have questions or comments? Email us at cyberleadersnetwork@sans.org