Artificial intelligence is simultaneously accelerating productivity gains across enterprises and fueling new security threats at unprecedented speed. In this new environment, organizations are turning to tools such as AI red team services to help strike a balance between innovation and protection.

CrowdStrike Holdings Inc. has been observing this transition over the past months. There’s been an obvious acceleration of AI adoption for internal use among many organizations, according to Tom Etheridge (pictured), chief global professional services officer of CrowdStrike.

“We’ve seen tremendous uptick in threat actor activity where AI is part of the portfolio of tools that threat actors are using to accelerate the speed at which they carry out their attacks,” he said. “In order to not only help from a product point of view, the services organization at CrowdStrike has made substantial investments in building offerings that enable customers on both fronts.”

Etheridge spoke with theCUBE’s Dave Vellante and Rebecca Knight at Fal.Con, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed the accelerating adoption of services such as AI red team within organizations and the growing use of AI by threat actors to intensify cyberattacks. (* Disclosure below.)

AI red team service

Among CrowdStrike’s offerings is an AI red team service to test AI tools for vulnerabilities by simulating adversary attacks. A lot of organizations have the capabilities to stand up infrastructure and start testing it, but they really don’t do it through a security lens, according to Etheridge.

“Running a red team operation against AI tooling can really help organizations understand where they might be at risk,” Etheridge said.

The goal of red teaming and other assessment services is to help customers understand the true scope of their AI security challenges. This need is being amplified by both a bottom-up surge of interest in AI tools and a top-down mandate to drive automation and efficiency across the business, according to Etheridge.

“That top-down push and the bottom-up push create this sprawl of AI usage that a lot of security and IT teams don’t really have great visibility into,” he said. “And as it’s being used, this is not just about tooling, it’s about data.”

Much of the focus centers on the data feeding and training AI models, which is drawn from across the business. In many cases, that data is highly sensitive, according to Etheridge.

“Running red team operations really helps understand what data is feeding these models,” he said. “Is it sensitive data? How are we insulating and protecting that information as we’re building out our AI tool structure?”

Security assessments often reveal hidden uses of AI that IT and security teams did not previously know about. In some cases, flawed inputs can manipulate a model into unintended actions, according to Etheridge.

“We really focus on getting organizations to understand how the AI models are structured, what kind of controls they have in place,” he said. “How are they governing their AI tooling, and then how are they making sure that it isn’t being misused by threat actors potentially?”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of Fal.Con:

(* Disclosure: TheCUBE is a paid media partner for Fal.Con. Neither CrowdStrike Holdings Inc., the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE