Why employment counts as high-risk under the EU AI Act
The EU AI Act treats AI systems used in employment and worker management, including recruitment, task allocation and performance evaluation, as high-risk. This does not ban their use. It means providers and users take on specific obligations around risk management, data quality, documentation, logging, transparency and human oversight before and during deployment.
- Recruitment and candidate screening tools
- Systems allocating tasks or shifts based on behaviour or performance
- Tools used in performance evaluation and monitoring
- Systems influencing promotion, termination or contract decisions
Under the EU AI Act, AI systems used in employment and worker management sit in the high-risk category, alongside areas such as critical infrastructure and access to essential services.
What high-risk does not mean
It is not a ban. It is a set of conditions that must be met before and during use, placing obligations on both the provider building the system and the organisation deploying it.
The obligations, in plain language
- 01Risk managementIdentify and document foreseeable risks the system could create for candidates or employees.
- 02Data governanceUse training and input data that is relevant, representative and checked for errors.
- 03Technical documentation and loggingKeep records that let the system's behaviour be reconstructed and reviewed.
- 04TransparencyGive affected people and operators enough information to understand how the system is used.
- 05Human oversightEnsure a person can meaningfully review, question and override outputs before they affect someone's employment.
For HR leaders, the practical starting point is an inventory: which tools touch recruitment, evaluation or task allocation, and which of these obligations are already in place.
Where to start
Begin with an inventory of AI-assisted HR decisions, then work through risk management, data governance and human oversight for each one, in that order.
Underlag
- Our assessment
AI systems used for recruitment, task allocation, and performance evaluation fall under high-risk categories in the EU AI Act
Common questions
- Does this apply to a vendor's off-the-shelf ATS?
- Likely yes if it screens or ranks candidates; the deploying organisation still carries obligations around oversight and use.
- Is a simple scheduling tool automatically high-risk?
- It depends on whether it materially affects work allocation or evaluation; a case-by-case assessment is needed rather than a blanket assumption.