AI risk is outpacing its value: Gartner

Sponsored by

Header: 59% of professionals spend 11+ hours a week chasing information across systems.†

Copy: This guide breaks down how to connect context across tools and keep work moving.

†Quickbase, 2025

It’s a world dominated by AI risks, and we’re just living in it.

During the opening keynote at the Gartner IT Symposium/Xpo in Orlando, Florida, Gartner Fellows Daryl Plummer and Kristin Moyer told the audience that AI risk is growing faster than AI value. (According to Gartner, 86% of CIOs are seeing AI risk outpace AI value in their organization, although the potential value is high.)

“In fact, it’s already compounding,” Moyer said. “Think of an interest-bearing account. With simple interest, the account grows linearly even if you do nothing. But with compound interest, money grows exponentially. Every AI risk compounds the risk that came before it.”

It’s all about balance. Plummer said CIOs need to align value with risk. Like drift car drivers, he added, companies need to balance maintaining the “right kind of momentum” with “avoiding the wrong kind of drag.”

“They don’t want momentum to be slowed by bad aerodynamics or underinflated tires,” Plummer said. “The drift cars do have brakes. We don’t put brakes on cars so they can go slow. We put brakes on cars so they can go fast, safely. Your brakes might be more strict AI usage policies with your data, or you might limit AI model access.”

Leveraging AI within software development is one example where AI value can potentially compound faster than AI risk, according to Plummer.

“These new AI tools, they create great momentum for software development,” Plummer said. “Generating massive amounts of code could be good, or it could be bad momentum, depending on the quality of code. You have to be careful.”

Where companies go wrong. Plummer added that value can only compound when “success builds on success” as opposed to adopting AI use cases for numbers’ sake: “Your task is to stop compounding volume and start compounding value.”

However, Plummer said vendors often push companies to chase “volume” as they promote their agents and AI services, adding that, while vendors pitch productivity gains of 50% or more, enterprises only see a 16% surge.

“Instead of monetizing your value or getting your returns, vendors want you to grow theirs,” Plummer said. “The biggest drag on AI momentum is using AI when you don’t need to.”

Moyer suggested that companies should shift away from tokenmaxxing to valuemaxxing, a strategy that measures AI by what it produces as opposed to how many tokens were used: “Start limiting careless consumption before it limits your critical use cases.”

Governance matters. Plummer and Moyer also said companies need to align safety with governance—something that very few leaders currently do. According to Gartner, only 25% of CIOs are aligning technology and business risk.

Prior to AI, Moyer said security risks were akin to having a dog: “You just had to leash them, fence them in, watch them, and scoop their poop.”

Now, with the rise of shadow AI, security risks mimic cats. “You wake up one morning, and they’re into everything,” Plummer said. “They’re in the laundry basket. They’re shredding the furniture and breaking into the snack cabinet.”

CISOs should treat shadow AI and related dangers like “crime prevention” as opposed to incident response.

“A CISO must actively monitor for misuse,” Plummer said. “Build an active threat defense using guardian agent technologies, because ongoing discovery must precede enforcement.”

Businesses should identify who owns AI platforms and tools, and reassess the stability of their security programs, Plummer continued: “Without a deep review and audit of your security policies, you’re just asking for trouble.”