just rewrite this section ‘How many people in your company use AI? Now, how many of them are using the AI platform your company actually approved? The reality is those numbers probably aren’t as aligned as one might hope. Today, we explore the burgeoning topic of visibility in IT, particularly how AI has seamlessly woven itself into the fabric of shadow IT. ‘

Understanding Shadow IT

Shadow IT isn’t new. If you’ve spent any time in IT, you know it isn’t always about rebellion or deliberately ignoring company policy. Usually, it’s just someone trying to get their work done. Procurement takes too long, security says no without offering an alternative, and eventually someone like our hypothetical Steve finds another way to get the job done.

Consider these classic examples:

• An Access database or Excel workbook suddenly becomes critical.
• Someone buys a NAS because the wait for new resources is endless.
• A basic Raspberry Pi is tucked under a desk running crucial apps.

“Shadow IT usually starts with frustration, not rebellion.“

Every department has its version of shadow IT, beginning not with malice but with necessity—a bid to meet deadlines infeasible by conventional means.

Feedback Loops: Learning from IT’s Past Mistakes

Understanding why shadow IT surfaces can reveal more about organizational inefficiencies than simple disdain for rogue processes. Look beyond the default assumption that shadow IT is merely an inconvenience. Often, it’s a signal flaring to highlight friction points in your existing processes.

When shadow IT arises:

• Is the root cause slow approval processes or outright rejection from procurement or security teams?
• Is there a missed business deadline that the official processes couldn’t address promptly?

Sometimes, shadow IT is less about technology and more about feedback that the official way of doing things isn’t working. IT departments are very good at keeping track of servers, hardware, firmware, and everything in the data center. The bigger risk may be somewhere nobody is looking: under someone’s desk or running in a browser.

AI Shadow IT: The New Invisible Risk

Shadow IT has evolved; AI is now the easiest tool to create it. In the past, rogue operations required physical resources—buying hardware, configuration, or clandestine administrative privileges. AI flips the script. With AI:

• You open a browser and type a prompt.
• No need for admin rights, installations, or procurement.

As a result, the barrier to entry for shadow IT is nearly nullified. Today, it’s as simple as querying ChatGPT or any AI tool for instant solutions, often involving sensitive company data.

Commonly Misused Data in AI Contexts:

• Contracts
• Customer emails
• Source code
• HR documents

Despite the benign intent, such practices can have dire consequences.

The Risks of Using AI in IT

AI’s answers often seem credible, which is exactly why they’re perilous. AI generates convincing responses—sophisticated, logical, and with polite apologies upon errors. It’s easy to trust, but unchecked reliance can lead to mishaps.

Imagine relying on AI for:

• Network configurations
• Firewall rules
• SQL queries

The risk manifests when AI fabricates a parameter or uses outdated commands, leading to system outages or data breaches.

One thing to remember: AI doesn’t know your environment. It doesn’t know your network, your configuration, or all the strange decisions that got you where you are today. Treat its answers accordingly. Check the backups, review the changes, and read the logs just as you always have.

AI Fragmentation: A New Kind of Chaos

The emergence of AI has splintered tool usage within organizations. While management may declare a single approved platform—perhaps Microsoft Copilot—the reality is different departments gravitate towards different tools:

• Marketing might favor ChatGPT.
• Developers might veer towards Claude.
• Design teams could lean into Midjourney, with sales exploring Gemini.

Each tool introduces its own challenges in terms of workflow ownership, testing, and maintenance. If the answer to who created or maintains it is “the AI did,” then we face a new breed of technical debt—one undocumented and precariously maintained.

How to Manage AI Shadow IT

Banning or blocking AI and shadow IT isn’t a sustainable strategy. We’ve tried this before with things like Dropbox and USB drives. Block the tool, and people who think they need it will eventually find another way.

Instead, more effective approaches include:

• Open Communication: Ask employees what tools they use. Often, this ten-minute conversation reveals more than months of policy drafting.
• Alternative Solutions: Offer easily accessible, secure alternatives. If the secure path is straightforward, employees will likely take it.

AI is less of a technology problem and more of a leadership problem. When people turn to unapproved tools, there’s usually a reason. Find the friction in the official process and fix it.

The goal isn’t to stop people from solving problems. It’s to give them a supported way to solve them without creating another system that IT only discovers six months later when something breaks.

Shadow IT has always filled gaps in processes, and AI doesn’t change that. It just makes creating new shadow IT much easier. Give people the right tools, sensible rules, and a path that actually works, and AI becomes something IT can support instead of something it has to chase.