AI in HR — Employment AI Governance and EU AI Act High-Risk Obligations

Definition

AI in HR (Human Resources) refers to the deployment of artificial intelligence systems in employment-related processes — including recruitment (CV screening, candidate ranking, interview analysis), performance management (automated evaluation, productivity monitoring), talent management (promotion recommendations, succession planning), workforce management (scheduling, task assignment), and employment decisions (termination risk scoring, compensation algorithms). Employment-related AI systems are among the most regulated under the EU AI Act, occupying a central position in the high-risk categories of Annex III.

HR AI systems create governance challenges across multiple dimensions. Algorithmic bias is particularly acute in employment contexts because historical hiring data reflects past discrimination patterns, and models trained on historical data can learn and amplify those patterns. Transparency and explanation requirements are strong because employment decisions are among the most consequential for individuals. Human oversight is mandatory under both the EU AI Act and good employment practice.

Beyond the EU AI Act, HR AI systems must comply with GDPR (employees are data subjects with full rights), anti-discrimination law (Employment Equality Directive, national anti-discrimination legislation), and in some jurisdictions, specific algorithmic employment regulations (New York City Local Law 144 on automated employment decision tools being the most prominent example globally).

Why it matters operationally

AI in HR matters because employment decisions are among the most consequential for individuals and among the most legally protected under anti-discrimination law. An AI system that systematically disadvantages candidates from specific demographic groups in hiring, promotion, or compensation creates discrimination liability independently of any AI-specific regulation — and the EU AI Act layers mandatory governance requirements on top of existing anti-discrimination obligations.

For organizations using AI in HR, the compliance landscape is clear: EU AI Act Annex III explicitly includes AI used in recruitment, selection, promotion, termination, and performance evaluation. These systems must undergo conformity assessments, maintain technical documentation, implement bias testing and human oversight, and register in the EU database for high-risk AI systems before deployment. Non-compliance carries fines of up to 3% of global annual turnover.

Regulatory framework

Framework	HR AI obligations
EU AI Act — Annex III	AI systems used to recruit or select persons, make promotion and termination decisions, evaluate performance, or monitor and assess persons in work contexts are high-risk systems.
GDPR	Employee data is personal data with full GDPR rights. Candidate profiling requires legal basis, DPIA if large-scale, and must comply with access and explanation rights.
Employment Equality Directive (EU)	Discrimination based on race, gender, age, religion, or sexual orientation in employment is illegal regardless of whether the cause is an algorithm or a person.
ISO/IEC 42001	HR AI systems within AIMS scope must be documented, risk-assessed, and subject to bias and human oversight controls.

How Zertia evaluates it

Zertia evaluates HR AI systems through the EU AI Act Assessment (classification, gap analysis, remediation roadmap for high-risk employment AI) and the High-Risk AI Systems Audit (conformity assessment evidence, bias testing, human oversight evaluation, technical documentation review). Zertia has particular experience with HR technology companies that provide AI-based employment systems to enterprise clients.

[EU AI Act Assessment] · High-Risk AI Systems Audit