Dutch Employees Embrace Artificial Intelligence Faster Than Their Companies Can Keep Up
Amsterdam, Sunday, 23 August 2026.
A new central bank study reveals only 25 percent of Dutch workers receive employer support for artificial intelligence, leaving most to use risky, unauthorized free tools.
The Rise of ‘Shadow AI’ in the Dutch Workplace
A profound shift is occurring across the Dutch professional landscape, characterized by a rapid phenomenon known as “Bring Your Own AI” (BYOAI) [1]. According to a study by De Nederlandsche Bank (DNB) published in the economic journal ESB, employees are integrating artificial intelligence into their daily workflows far more rapidly than their organizations are establishing policies to govern them [1][2]. The central bank’s survey, originally conducted last October, revealed that a mere 25 percent of Dutch workers feel their employers actively support or encourage the use of AI at work [1][2]. This leaves the vast majority of employees to navigate these advanced technologies on their own, often logging into consumer-grade versions of platforms like ChatGPT or Claude using personal accounts on corporate networks [1].
The Security Risks of Unmonitored Adoption
This disconnect between rapid individual adoption and lagging corporate oversight has created a severe governance blind spot for businesses [1]. Because workers lack official company rollouts and structured guidance, they frequently input proprietary corporate data, client information, or sensitive source code into public models without realizing they are forfeiting data privacy [1]. This is not a behavioral or discipline problem of the employees themselves, but rather an organizational failure to track where sensitive data is being transmitted [2]. Without clear policies or enterprise-grade software, companies remain completely unaware of which AI tools are being utilized on their networks from week to week [2].
Sectoral Disconnects and Bureaucratic Hurdles
The gap between AI usage and employer support is visible across all sectors, but it is most pronounced in highly regulated areas [1][2]. In the public sector and government, 51 percent of workers use AI, but only 22 percent feel their employers support it [2], representing a gap of 29 percentage points. In the education sector, the gap is also substantial: 63 percent of workers use AI tools, while only 36 percent report receiving active support [2], leaving a gap of 27 percentage points. In healthcare (zorg), 37 percent of employees utilize AI, but a mere 15 percent experience employer backing [2], a gap of 22 percentage points.
Regulatory Paralysis and the Licensing Divide
The substantial gaps in education and government, which approach or exceed 30 percentage points, are rooted in strict bureaucratic regulations and data protection laws such as the General Data Protection Regulation (GDPR) [1]. While teachers and civil servants use generative tools to rapidly draft lessons, summaries, or reports, institutional policies remain paralyzed by legal uncertainties regarding copyright, accountability, and student or citizen data safety [1]. In contrast, sectors like IT, finance, and professional services show much smaller gaps because companies are aggressively purchasing enterprise-grade, closed-loop AI software licenses like Copilot for Microsoft 365 to capture immediate billable productivity gains [1]. Nationwide, only three out of ten AI users work with a paid license—primarily those who receive explicit employer support—while the remaining 7 out of ten users rely on free, unsecured versions [2].
The Paradox of Unstructured Productivity
While the adoption of generative AI is driven by a desire for efficiency, the lack of structured corporate guidance has created an unexpected paradox on the ground. Recent studies reveal that over two-thirds of workers using generative AI save between three and eight hours per week [1]. However, this time savings rarely translates into a lighter workload [1]. Instead, because employers have not established clear pathways or adjusted expectations, the saved time is immediately filled with a higher volume of tasks [1]. This compresses deadlines and increases corporate speed expectations, leaving workers feeling more rushed and overwhelmed than they did before they began using AI [1].