Shadow AI Is a Leadership Problem, Not a Security Problem

Companies bought AI tools in 2025 and are now asking what those tools are actually doing. Meanwhile, employees are already using AI on personal devices and sometimes for work, without telling anyone. The gap between what organizations demand from AI and the safety they provide for people to learn it is where adoption stalls and the most capable employees start looking elsewhere.

Bethanie Nonami is Founder and CEO of VINify, a vehicle surveillance intelligence company. She is also a certified AI consultant and founding member of the International Association of AI Consultants. Her career spans over 30 years in technology and enterprise AI, including leadership roles at IBM and Lodestar Solutions.

"People are using ChatGPT on their phones, or they're using it at home, or they might even be using it for work, but they're not telling their organization because they don't know if they look bad or they look good. It's developed a whole term and almost a stigma around how we navigate this," says Nonami.

No playbook means no permission

Nonami frames the leadership problem directly: organizations are telling leaders to implement AI without giving them guidance on what that looks like. The result is tool purchases without direction.

"Some organizations are like, 'Leader, implement AI, but I'm not going to put guardrails around what that looks like,'" Nonami says. "So what do we do? We buy tools, we turn them on, and we say go be awesome." Without clear framing, employees default to hiding their use rather than sharing what they are learning.

"We're rewarded when we have the right answer, all the way going back to school," Nonami says. "Vulnerability isn't a leadership trait that some organizations allow." In cultures where executives cannot appear uncertain, the signal to employees is clear: figure it out quietly.

Two weeks, not six months

Nonami argues that traditional change management cycles are too slow for a technology that updates weekly. Her prescription is to compress experimentation until it no longer feels like a bet.

"A lot of people bought stuff in 2025 and they're asking, 'What is this AI thing doing?'" Nonami says. "What's one thing we can solve? Let's try and use AI to figure it out in two weeks. Does it save us time? Does it give us more work? Is it helping us?" Two-week cycles give leadership enough visibility to stay comfortable while giving teams enough room to learn without career risk.

"This is different than learning Microsoft Word or navigating the Internet," Nonami says. "There's so much psychology and risk and fear around it that everything has shifted." The technology changes faster than any training program can keep up with, which means the learning model itself has to change.

Office hours and communities of practice

Nonami describes a practical model she has seen work in organizations that take the culture problem seriously.

"We've seen organizations do office hours where they're open to talk about AI. Bring your problems, bring your wins, bring your lessons," Nonami says. "It's crickets at first. Nobody shows up. And then people start, and they learn something, and it grows." She extends this to communities of practice built on existing employee resource groups where people already feel safe. "Give them a safe place where they can fail without feeling judged from a performance standpoint."

Organizations invest heavily in permissioning for agents and data access, but there is no equivalent investment in permissioning for people to explore and fail. "Set the guardrails, tell them what they can and can't do," Nonami says. "But give them the option to ideate and figure it out. We've got to try something that may be uncomfortable."

The talent stakes are real

Nonami closes on the competitive consequence. AI skills have increased market value and salary demands to levels that give employees real leverage. A finance professional or legal specialist with AI fluency can command a 30% increase by moving to an organization that supports their growth. "If your people are trying to help your organization grow and you don't give them what they need to support that, why would they stay?" Nonami says.

The risk for companies that cannot create these environments is not just slower innovation. It is that competitors who build cultures around AI learning will attract the talent that everyone else loses. "Behind is such a buzzy word right now," Nonami says. "But it's going to be harder and harder to catch up the further and the slower you move."