AI transformation in most organizations follows a familiar pattern. Management decides that they “need to do something with AI.” A steering committee is convened. A pilot is set up. Then something happens that isn’t on any roadmap: employees are already where management is only just trying to go.
A thesis at the FOM that I supervised provides empirical evidence for a pattern I regularly see in consulting engagements: there is a gap between AI usage among the workforce and guidance from the leadership level. And it is precisely this gap that becomes a compliance issue.
The Gap Between Usage and Guidance
A 2025 Boston Consulting Group study shows that 72 percent of employees use AI applications regularly. A McKinsey study adds that three times as many employees as executives suspect report that AI is already taking over more than 30 percent of their daily work.
Only a quarter of employees experience clear support from their managers. Furthermore, the thesis I supervised shows a surprising finding: whether someone held a leadership role had no significant effect on the fulfillment of basic psychological needs such as autonomy, competence, and social relatedness within the sample.
According to Rogers’ Diffusion of Innovation theory, the opposite would be expected, as managers are seen as early adopters and opinion leaders. However, the paper contextualizes this finding: the sample consisted predominantly of younger respondents who grew up with digital technologies and is not representative, so the result cannot be generalized.
The decisive distinction is therefore less about the formal leadership title and more about whether leadership actively supports the use of AI.
Where the Gap Becomes Dangerous: Shadow AI
This leadership gap has a direct consequence that is becoming increasingly visible in auditing and compliance: Shadow AI. More than half of the respondents in the BCG study state that they also use unauthorized AI tools.
This is not a matter of intentional rule-breaking. Employees use AI because they want to be more productive. If their organization does not provide official channels and clear rules, they seek unofficial ones. The risks are significant: GDPR violations during data transfer, regulatory gaps regarding the EU AI Act, and operational security vulnerabilities due to a lack of transparency.
3 Levers Against Shadow AI
The thesis suggests focusing on the interplay of three elements:
- Clear governance that is practical for everyday use. Which tools are allowed? For which data? With which approvals? These questions need clear answers that employees can implement in their daily routines. Compliance architecture that is overwhelming will be bypassed.
- Active empowerment instead of passive permission. When employees receive official tools along with real training, the incentive for Shadow AI decreases. The experience of competence is the strongest lever for AI acceptance. Those who empower win employees over to the official channels.
- Executives as visible role models. 77 percent of respondents in the BCG study state that they perceive AI as more valuable when executives actively support its use. Leadership visibility is not a communication issue, but an acceptance lever.
Conclusion: Leadership as a Dual Requirement
AI transformation without an active leadership role is not a transformation. It is an uncontrolled movement that leads to compliance risks.
The data clearly shows: the employees are there. The technology is there. What is missing is the leadership level to bring it all into a form that is both regulatorily viable and humanly accessible. This dual requirement is the actual leadership task of the coming years.
Scientific background
This article draws on findings from the following work:
Gegaj, A. (2026): Perception and Evaluation of Artificial Intelligence – the Influence of Leadership and Motivation. Thesis, FOM University of Applied Sciences.
Other cited studies: Boston Consulting Group (Beauchene et al. 2025), McKinsey (Meyer et al. 2025).
