Uploaded November 2026 | Updated November 2026, 13 minutes ago
Check out Palo Alto Networks here: https://ul.live/Cortex-Cloud_yt
I sat down with Prajakta Jagdale, VP of InfoSec at Palo Alto Networks, who got early access to Anthropic's Mythos Preview and watched his team realize how capable it was at offensive security. We talked about what happens when finding vulns stops being the hard part: corporate friction, patching every version at once, auto-enforcement, and why the answer keeps coming back to fundamentals.
What we talk about:
Who has the edge: Why PJ thinks attackers still have a slight advantage, even with equal access to frontier models, and how much the same tools speed up triage, PSIRT, and patch validation for defenders.
Friction is the real enemy: My take that the biggest danger to defenders is corporate politics and bureaucracy, not technology, and how a POC written by a frontier model gets engineering to take a finding seriously.
Fixing at the speed of finding: Secure base images that cut the unpatched-CVE backlog by 80%, and AI pipelines that brought open-source supply chain response from days down to two or three hours.
Going nuclear on the backlog: Auto-enforcement that shuts down cloud resources without Cortex XDR and firewalls off anything exposed to the internet, plus why a deadline is what finally gets engineering moving.
AI in the development lifecycle: Validating code-scan findings by exploiting them, tuning scanning rules based on what frontier models keep exploiting, and harnesses that pen test builds when they deploy.
00:00 - Introduction and early access to Mythos Preview.
01:46 - Do attackers or defenders have the advantage?
03:50 - Corporate friction as the biggest danger for defenders.
06:33 - Building trust with engineering and patching every version.
09:06 - Splitting investment between hygiene and detection and response.
11:34 - Golden images, secure base images, and supply chain response.
15:50 - Deception and canaries.
16:41 - Back to fundamentals and auto-enforcement.
19:51 - Getting visibility into agentic threats.
20:31 - Securing the development lifecycle with frontier AI.
21:48 - How InfoSec partners with Palo Alto's product teams.
23:48 - How the team stays ahead.
25:55 - Where to learn more.
Subscribe to the newsletter at: danielmiessler.com/subscribe
Join the UL community at: danielmiessler.com/upgrade
Follow on X: https://x.com/danielmiessler
Follow on LinkedIn: linkedin.com/in/danielmiessler
Check out Palo Alto Networks here: https://ul.live/Cortex-Cloud_yt
I sat down with Prajakta Jagdale, VP of InfoSec at Palo Alto Networks, who got early access to Anthropic's Mythos Preview and watched his team realize how capable it was at offensive security. We talked about what happens when finding vulns stops being the hard part: corporate friction, patching every version at once, auto-enforcement, and why the answer keeps coming back to fundamentals.
What we talk about:
Who has the edge: Why PJ thinks attackers still have a slight advantage, even with equal access to frontier models, and how much the same tools speed up triage, PSIRT, and patch validation for defenders.
Friction is the real enemy: My take that the biggest danger to defenders is corporate politics and bureaucracy, not technology, and how a POC written by a frontier model gets engineering to take a finding seriously.
Fixing at the speed of finding: Secure base images that cut the unpatched-CVE backlog by 80%, and AI pipelines that brought open-source supply chain response from days down to two or three hours.
Going nuclear on the backlog: Auto-enforcement that shuts down cloud resources without Cortex XDR and firewalls off anything exposed to the internet, plus why a deadline is what finally gets engineering moving.
AI in the development lifecycle: Validating code-scan findings by exploiting them, tuning scanning rules based on what frontier models keep exploiting, and harnesses that pen test builds when they deploy.
00:00 - Introduction and early access to Mythos Preview.
01:46 - Do attackers or defenders have the advantage?
03:50 - Corporate friction as the biggest danger for defenders.
06:33 - Building trust with engineering and patching every version.
09:06 - Splitting investment between hygiene and detection and response.
11:34 - Golden images, secure base images, and supply chain response.
15:50 - Deception and canaries.
16:41 - Back to fundamentals and auto-enforcement.
19:51 - Getting visibility into agentic threats.
20:31 - Securing the development lifecycle with frontier AI.
21:48 - How InfoSec partners with Palo Alto's product teams.
23:48 - How the team stays ahead.
25:55 - Where to learn more.
Subscribe to the newsletter at: danielmiessler.com/subscribe
Join the UL community at: danielmiessler.com/upgrade
Follow on X: https://x.com/danielmiessler
Follow on LinkedIn: linkedin.com/in/danielmiessler










