The European AI Act does not only regulate technology in the classical sense. As soon as an AI system has an influence on the employment relationship — for example by assigning tasks, allocating work, evaluating performance, or playing a role in hiring or dismissal — that system falls under the specific attention of the regulation. The precise classification and obligations depend on the risk level assigned to the application. For the current, legally binding text and the annexes with risk categories, we refer to the official publication of the regulation in the Official Journal of the European Union and to the implementing guidelines that the European Commission publishes in relation to it.
This page describes the mechanism: when the use of AI becomes relevant in a personnel context, what role the employer gets in that, and where you should turn for the precise obligations. It is not advice about your specific situation and not a basis for a personnel decision.
The regulation looks at the function of an AI system, not at the name of the software. A planning tool, a scheduling application, or a system that allocates work can fall under the rules as soon as it has an effect on the position of employees. Examples of functions mentioned in the context of work are systems that:
Whether a specific system is in practice classified as high risk depends on the precise design and use. A more extensive explanation of this classification and the associated obligations can be found on the page about high-risk AI in the workplace.
The regulation distinguishes different roles: the provider that develops or places the AI system on the market, and the deployer (the party that puts the system to use in practice). An employer that acquires and deploys a planning or evaluation system usually acts as deployer. That role brings its own obligations, such as informing those concerned, ensuring sufficient human oversight, and keeping certain records about the use. The precise scope of these obligations differs per risk category and is set out in the regulation itself and in the guidelines published alongside it.
In addition, there is a separate obligation, distinct from the AI Act, concerning the level of knowledge of staff who work with AI systems or make decisions about them. What that obligation entails and to whom it applies can be read on the page about AI literacy as an obligation.
This page addresses the question of whether and how the AI Act applies to a system that directs or evaluates work. It is not an explanation of employment law procedures. Dismissal, reorganization, changes to job content, or works council involvement when introducing new systems have their own statutory requirements, among others from labor law and the Works Councils Act. Those requirements are separate from the AI Act and are not addressed here. A decision about personnel requires its own legal advice; this page does not provide a basis for that.
The AI Act regulates the conditions under which an AI system may be deployed. It says nothing about which part of the work in a role can actually be automated. That is a factual question about tasks: which parts of a role consist of repeatable, well-defined steps, and which parts require judgment, context, or responsibility that remains with a human. That question plays out differently across sectors — the trade-off in what AI can take over in construction differs from that in what AI can take over in healthcare, and even within what AI can take over in the installation sector, tasks vary greatly in the degree to which they lend themselves to automation.
We map that factual question using three categories: tasks that AI can take over, tasks where AI makes a proposal and a human approves or rejects it with reason, and tasks that remain human work. That classification says something about the task, not about the legal status of the system being used, and not about what an employer does with the outcome.
If you want to know whether and how the AI Act applies to a system you use or are considering, the most direct step is to consult the regulation and the accompanying guidelines, or to have this assessed by an advisor with knowledge of the AI Act and employment law. This knowledge base does not provide a substitute for that.
If you would like a first, factual picture of which part of the tasks in your job profile can be automated today, you can fill in the free quickscan from ftetoai: twelve questions, no account required, resulting in an indication of the share of hours that can be taken over by AI today. This is an indication based on tasks, not a legal assessment and not personnel advice. The full work scan, with a more detailed analysis per role, is still under construction.
Vraag maar. Ik ken de kennisbank van deze site; wat ik niet weet, zeg ik erbij.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.