any city park and you will notice narrow dirt trails cutting across the grass. They appear between sidewalks, across lawns, and through corners planners never intended people to cross.
Urban designers call these desire paths.
They form when people choose their own routes instead of the official walkways. Over time the grass disappears and the informal trail becomes visible evidence of how people actually move through a space.
For decades, planners treated these paths as mistakes. Today many see them differently. Desire paths reveal something valuable. They show where the original design failed to match human behavior.
Something similar is happening inside modern organizations.
Employees are already using artificial intelligence to draft emails, analyze data, summarize documents, and generate ideas. A marketing manager may use a language model to prepare campaign copy. A finance analyst may summarize reports with an AI assistant. A product manager may test ideas through generative tools.
Often this experimentation happens quietly, outside official systems or policies.
This phenomenon has a name: Shadow AI.
The term echoes the older concept of shadow IT, when employees installed software without approval from corporate IT departments. Today the pattern is repeating itself with artificial intelligence. Workers bring generative tools into their daily workflows long before organizations establish governance structures or approved platforms.
This raises obvious concerns. Sensitive corporate information can enter external systems without clear visibility into how that data is processed or stored. Regulatory frameworks such as GDPR or the EU AI Act may be violated unintentionally. Security teams lose oversight of how information moves through the organization.
Yet focusing only on risk misses something important.
Shadow AI often reveals where existing systems are no longer keeping pace with how people need to work. Like desire paths in a park, Shadow AI exposes where employees are searching for faster and more intelligent ways to complete everyday tasks.
If this behavior were rare it might be manageable. The numbers suggest otherwise.
Surveys indicate that nearly four out of five people using AI at work bring their own tools rather than relying on systems provided by their employer. Many interact with these tools through personal accounts instead of enterprise platforms designed to protect sensitive data.
The consequences are beginning to surface. Studies suggest that more than half of employees admit to entering confidential information into AI systems. Organizations experiencing widespread Shadow AI usage report higher breach costs and greater exposure to regulatory risk.
In other words, artificial intelligence is already spreading through workplaces at scale. Governance, training, and security frameworks are arriving later.
This gap creates real risks. It also reveals something about how technological change actually unfolds inside organizations.
Shadow AI as an organizational signal
There is another way to interpret Shadow AI.
When employees adopt new tools outside official channels they are not only bypassing governance structures. They are also revealing where existing workflows are failing them.
In many organizations, generative AI appears first at the margins of daily work. Employees experiment with drafting emails faster, summarizing documents, analyzing spreadsheets, preparing presentations, or exploring ideas. These experiments happen quietly because the official systems available to them do not yet support these capabilities.
What security teams see as unauthorized usage can therefore function as a form of organizational diagnostic. Shadow AI reveals where people are trying to move faster than the systems around them allow.
Urban thinkers have long observed a similar pattern in cities. Jane Jacobs argued that cities should be designed around how people actually move through them, not around how planners imagine they should. The informal paths across parks and campuses provide a map of real behavior.
Organizations facing the rise of Shadow AI may need to adopt the same mindset.
Instead of viewing Shadow AI only as a governance failure, leaders can treat it as an early signal of where artificial intelligence might deliver the greatest value. The informal experiments appearing across teams often point to workflows where automation, augmentation, or improved access to information could significantly increase productivity.
When organizations approach these patterns with curiosity rather than fear, the scattered experiments begin to reveal something valuable. They highlight repetitive tasks employees are already trying to accelerate and expose processes where better tools could unlock meaningful efficiency gains.
What first appears chaotic often points to opportunities for consolidation. Instead of dozens of fragmented experiments across departments, organizations can identify common needs and build governed, scalable solutions around them.
Handled well, this shift does more than reduce risk. It empowers employees with secure tools that support the way they already work, turning artificial intelligence from something that requires constant supervision into a multiplier of creativity and innovation. Ignoring Shadow AI means missing these signals. It allows costly and uncoordinated experiments to continue in the shadows while organizations overlook insights that could guide smarter adoption.
Learning from the AI footpaths
Organizations that want to govern artificial intelligence effectively must first understand how it is already being used.
Shadow AI should not only be investigated as a compliance problem. It should be examined as a signal of where employees are attempting to move faster than the systems around them allow. The first step is visibility. Leaders need to understand which tools employees are already using and why. Employee surveys, technical audits, and open discussions across departments often reveal where experimentation is happening first. Marketing, sales, finance, HR, and product teams frequently emerge as early adopters.
Once these patterns become visible the challenge shifts from suppression to structure. Organizations must define which tools are appropriate, establish governance policies aligned with data sensitivity and regulation, and design processes that reflect how work actually happens inside the organization.
Culture matters just as much as policy. Employees should feel safe discussing how they are experimenting with artificial intelligence rather than hiding it. When people fear punishment or additional workload for adopting new tools, experimentation does not disappear. It simply moves further into the shadows.
Effective governance therefore requires more than rules. It requires an environment where responsible experimentation is encouraged and guided. Training, access to approved tools, and clear guardrails allow organizations to transform scattered experiments into coordinated progress.
Understanding what already exists in the shadows is often the first step toward building a resilient and intelligent AI strategy.
A final thought
In practice, Shadow AI is rarely the result of malice. More often it reflects misalignment and a lack of communication inside the organization. When employees feel unsafe sharing their experiments, when curiosity is met primarily with correction, the predictable outcome is silence.
People do not stop experimenting. They simply stop sharing.
If organizations want to govern AI effectively, they must begin by creating environments where thoughtful exploration is possible. Training, practical examples, and clear guardrails make responsible experimentation visible instead of hidden.
But culture matters most. When curiosity replaces suspicion, experimentation moves out of the shadows and into the open.
The first step toward governing Shadow AI is simple: understand where people are already walking.
About Aleksandra Osipova
Aleksandra Osipova is the founder of Apricity Lab, where she works with leaders and organizations navigating the transition toward AI-enabled systems.
She writes about artificial intelligence, systems thinking, and the future of work. More of her work and insights can be found on her LinkedIn.
