PodcastNo Priors

The New Apex Predator: CrowdStrike CEO George Kurtz on Adversaries Armed With AI Agents

Written by an Ethmos research agent · shared with you

CrowdStrike is racing to match attacker-grade AI with its own NVIDIA-built frontier security models as breakout times shrink to seconds.

ShowNo Priors

Episode dateOctober 22, 2026

Runtime45 min

FeaturingSarah Guo, Elad Gil

Agent state flattens the threat pyramid: Kurtz says ablated open-weight models give hacktivists nation-state-level capability, making the "agent state" the new apex predator.
Breakout times collapse: Average adversary breakout time is now 29 minutes, with the fastest observed at 27 seconds, driven by inference-speed prompt engineering rather than malware signatures.
CrowdStrike builds its own frontier lab: Working with NVIDIA on Nemotron, CrowdStrike trained red (Red Tempest) and blue (Blue Solano) security models under the SafeMind/SafeBind system, with a digital-twin environment for continuous red-vs-blue iteration.
Cost and data moat: CrowdStrike's post-trained security model costs about 3 cents per remediation vs. roughly $10 for top frontier models, backed by over a decade of proprietary threat data.
Hugging Face lesson: Attackers bypassed refusals on frontier models by switching to open-weight models, revealing defenders lack equivalent frontier-grade AI tools.
Platform expansion and optimism: CrowdStrike now runs 33 product modules (up from one), and Kurtz argues AI, like past tech shifts, will net-create jobs and make security an enabler rather than a brake on enterprise AI adoption.

Deep dive

The agent state as apex predator

Kurtz frames the shift using his own "pyramid of pain" model: hacktivists at the base (least skill, most numerous), mid-tier criminal groups, and nation-states at the top (most sophisticated, fewest actors). He says ablated, guardrail-stripped models have flattened this pyramid entirely — a hacktivist today can wield the tradecraft a nation-state possessed just a few years ago. The result, per Kurtz, is a barrage of faster, more sophisticated attacks where it's "not malware anymore, it's prompts" — attackers engineer prompts that generate unique signatures on each pivot, making detection harder. He notes that given enough time, an agent tasked with breaching a system will eventually find a way in.

The full analysis is a free trial away.

An Ethmos agent found this conversation, listened, and wrote this brief. Yours can do the same — across the shows you follow and the thousands you don't.

Read free — start your trial

TMTB readers: 14-day free trial + 25% off your first month

Also inside: 01 Open-weight models and the attack toolchain · 02 Geography and attacker profile · 03 CrowdStrike's frontier lab and platform build-out · 04 Differentiation, distillation, and defense-dominance case

This page is an AI-generated summary and analysis prepared with Ethmos and shared by an Ethmos user. It does not reproduce the original programming. The underlying episode and its recording remain the property of their respective creators, and all show and company names are the property of their respective owners.

AI summary. May contain errors. Not investment advice.