Skip to content
No PriorsNo Priors

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

Adversaries are already running AI agents, and the fastest attacks now play out in seconds. Sarah Guo and Elad Gil sit down with CrowdStrike president and CEO George Kurtz on what AI-equipped attackers mean for enterprise security. George explains how agents are flattening the adversary pyramid by putting advanced tradecraft in the hands of low-skill attackers, and why CrowdStrike and Nvidia are building a frontier-style lab of their own to train security models. They also get into how open-weight and closed models compare on security, where CrowdStrike is directing its platform investment, lessons from the recent Hugging Face incident, and why George sees AI, like every platform shift before it, as a net creator of opportunity. Sign up for new podcasts every week. Email feedback to show@no-priors.com Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil | @CrowdStrike | @George_Kurtz Chapters: 00:42 – Trailer 00:59 – George Kurtz Introduction 00:12 – Agentic Landscape 01:34 – AI and Adversaries 03:40 – Open Weight Models and Adversarial Attacks 05:27 – Adversary Geography 07:24 – Attacker Capability in Context 10:14 – Hugging Face Takeaways 12:49 – CrowdStrike Frontier Security Models 15:22 – CrowdStrike Key Offerings 18:41 – What Differentiates CrowdStrike’s Models 22:47 – Near Term State of Security World 24:37 – State of Model Distillation 26:42 – Open Weight Model Importance and Cost 28:49 – Assessing Real Agent Threat 31:29 – Areas to Watch in AI 33:52 – Case for Defense Dominance 38:03 – Executive Advice 42:05 – Shifts Create Opportunity 44:14 – Conclusion

George KurtzguestSarah GuohostElad Gilhost
Oct 22, 202645mWatch on YouTube ↗

At a glance

WHAT IT’S REALLY ABOUT

AI agents turn cybersecurity into a speed war for defenders

  1. George Kurtz argues AI agents have reshaped cybersecurity by giving low-sophistication attackers near–nation-state capability, increasing both the speed and volume of intrusions.
  2. He claims attackers increasingly use open-weight models because frontier models refuse many cyber-related tasks, enabling powerful offensive workflows to run locally without guardrails.
  3. CrowdStrike is focusing on identifying and instrumenting enterprise agents end-to-end—discovering shadow AI, tracking tool calls and identities, and enforcing policy and blocking rules.
  4. To keep pace with AI-driven offense, Kurtz advocates “fighting AI with better AI,” including red/blue security models trained on proprietary incident data and tested in a customer-specific digital-twin loop.
  5. He predicts growing enterprise reliance on cheaper open-weight models (with selective routing to frontier models for highest-value tasks), alongside continued model distillation pressure and increased importance of trust and AI sovereignty.

IDEAS WORTH REMEMBERING

5 ideas

AI agents flatten the attacker “pyramid,” creating an “agent state” apex predator.

Kurtz argues that readily available AI agents and cyber-focused open-weight models have collapsed the traditional gap between low-skill attackers and nation-state capabilities, increasing both attack volume and sophistication. This shifts the baseline threat model for every enterprise.

Open-weight models are a major accelerant for real-world offensive cyber activity.

He claims open-weight models are often preferred by attackers because frontier models’ safety refusals block many cyber-related requests, while open-weight models can be run locally without guardrails. This makes powerful offensive capability cheap and widely accessible.

Breakout time is collapsing—attackers pivot in minutes (or seconds), not days.

CrowdStrike measures how quickly an intruder can move from an initial foothold to more valuable systems; Kurtz says that timeline has shrunk from hours/days to minutes, with extreme cases in seconds. He links this speed to “inference speed” and rapid iteration enabled by AI.

Detection is getting harder because attacks can be ‘prompt-shaped’ and uniquely iterated.

Rather than relying on known malware signatures, attackers can generate unique prompt sequences and tool-use paths that change behavior run-to-run. Kurtz frames this as a shift from “malware” to “prompts,” complicating detection based on static indicators.

Shadow AI is now a core security problem: organizations don’t know what agents exist or what identity they run under.

He describes enterprises having many more agents operating than they realize, often with unclear identities/permissions and minimal oversight. The core defensive need becomes discovering agents, tracking their tool calls and sub-agents, and enforcing policy from “birth to behavior.”

WORDS WORTH SAVING

5 quotes

The new apex predator is the agent state. Not the nation state, it's the agent state.

— George Kurtz

What people have to keep in mind is it-it's not malware anymore, it's prompts.

— George Kurtz

For the first time, security is the gas pedal, not the brake.

— George Kurtz

Trust is, is, uh, built in drips and lost in buckets.

— George Kurtz

The only way to fight AI today is with better AI.

— George Kurtz

Agentic threat landscape and “agent state” framingShadow AI discovery and agent identity controlOpen-weight vs frontier models and guardrails/refusalsBreakout time, speed of inference, and rapid attacker iterationHugging Face agentic attack lessonsCrowdStrike SafeMine, Red Tempest, Blue Solano, and digital twinsAI sovereignty, trust, and model cost economics

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

Read along on your phone, search it beside the video, or ask your favorite AI app what was said.

iOSAndroidClaudeChromeChatGPTComing soon