Enterprises are rapidly adopting artificial intelligence, but firms are struggling to match deployment with strong AI agent guardrails, from clear ownership to secure-by-design architectures, to keep maturing agents within acceptable risk boundaries.

Recent research suggests that the spread of AI agents is swelling the ranks of non-human identities, widening the potential attack surface and making recovery harder. Of course, no control is ever impermeable, but better detection and prevention are achievable — and increasingly essential, according to Kavitha Mariappan (pictured, right), chief transformation officer of Rubrik Inc.

“AI introduces so much potential and opportunity. At the same time, unmanaged and ungoverned, it introduces and amplifies the level of risk within an organization,” Mariappan said. “Whether it is … managing and protecting against AI-generated or accelerated threats, or … safely deploying AI for innovation initiatives and things like that, organizations are having to start to think about what that looks like.”

Mariappan spoke with Michael Ortega (left), director of AI marketing at Rubrik, during theCUBE’s coverage of the Rubrik “Resilience for Everything: Cloud, Identity, AI” interview series. They discussed the role of agentic resilience in preparation for a new wave of AI-powered attackers. (* Disclosure below.)

AI agent guardrails facing mounting pressure

Rogue agentic AI has moved from science fiction to a boardroom item on the risk register. Rubrik’s own research found that a majority of IT and security decision-makers expect that, within the next year, at least half of the cyberattacks they face will be driven by agentic AI, Mariappan noted.

“We’re just at the tip of the spear right now, tip of the iceberg,” Mariappan said. “We’re going to see so much more of this.”

To stay ahead, Rubrik is helping enterprises catch up by providing them with dedicated control layers for AI agents, transforming policy into live control. That control layer stays active in production, logging and auditing in real time, Mariappan explained. But another ambitious bet is on reversibility.

“I would say what’s even more exciting is Agent Rewind, that gives us — and it gives organizations — the ability to actually roll back any sort of agentic actions that would have been inadvertent, unplanned or malicious, to a clean state.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of the Rubrik “Resilience for Everything: Cloud, Identity, AI” interview series:

(* Disclosure: Rubrik sponsored this segment of theCUBE. Neither Rubrik nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE