What happens when software starts thinking and acting like a human security analyst, only faster and without sleeping? That’s the reality cybersecurity leaders now face as artificial intelligence, particularly in the form of agentic AI, accelerates a shift toward autonomous threat response and real-time decision-making.

These digital superusers are changing the game. Capable of scanning vast datasets, triggering automated workflows and uncovering vulnerabilities in seconds, agentic systems are forcing organizations to rethink everything from data governance to platform strategy. With adversaries adopting the same tools and the lines between cybercriminals and nation-states increasingly blurred, the urgency to stay ahead has never been more pressing, according to George Kurtz (pictured), founder, president and chief executive officer of CrowdStrike Inc.

CrowdStrike’s George Kurtz talks about agentic AI with theCUBE.

“We’re still in the early innings, and I think everyone’s trying to figure out how they can use it, how they can control the data, how they get the results in a secure way,” Kurtz said. “To be honest, in a lot of areas, what’s keeping people up at night is just the Wild Wild West of AI everywhere … it’s got a lot of people staying up late at night.”

Kurtz spoke with theCUBE’s Dave Vellante at the RSAC 2025 Conference, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed how artificial intelligence, specifically agentic AI, is transforming cybersecurity by enabling autonomous threat detection and response while introducing new risks that demand adaptive, data-driven strategies. (* Disclosure below.)

Agentic AI creates new challenges and new opportunities

Unlike earlier chatbots or basic copilots, agentic AI systems are designed to take independent action based on policies, workflows and real-time data. CrowdStrike’s Charlotte AI goes beyond prompting to deliver automated incident response, threat investigation and operational outcomes with minimal human input, according to Kurtz.

“We were thinking beyond chatbots, wiring it into a workflow and getting real outcomes for customers,” he added. “You can have these sort of AI agents do work on your behalf, so they can start proactively researching and pulling information together; [they can] start proactively looking at incidents and summarizing those, and kind of figuring out what an analyst would do without necessarily all the prompting.”

This new generation of AI also creates risks, especially as adversaries adopt the same tools. Generative AI and large language models are helping threat actors rapidly identify and weaponize exploits, forcing defenders to adapt at unprecedented speeds, Kurtz explained.

“You can essentially use a lot of these different LLMs to help you create new attacks,” he said. “They can be used to understand where vulnerabilities are. They can actually create and weaponize new exploits. What we’re seeing is the window of time a company has to be able to sort of patch these zero days is dramatically shrinking. Security really is a function of time.”

As attackers become more sophisticated, platforms must evolve to stay ahead. CrowdStrike’s answer is a curated, open approach that emphasizes high-fidelity data and partner integrations. Unlike vendors pushing closed mega-platforms, Kurtz argued that CrowdStrike’s strategy focuses on strengths and interoperability.

“We don’t necessarily subscribe to that approach, because in security… there isn’t one platform to rule the world. It just isn’t,” Kurtz said. “But we can consume data from these other third parties, and we can combine that with our first-party CrowdStrike data, and then we can create a seamless experience across the platform in terms of hunting, in terms of AI, in terms of outcomes.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of the RSAC 2025 Conference event:

(* Disclosure: TheCUBE is a paid media partner for the RSAC 2025 Conference. The sponsors of theCUBE’s event coverage do not have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE