Article 13 Transparency for HR Tech AI — EU AI Act

How EU AI Act Article 13 transparency applies to HR and recruitment AI — candidate notification, scoring explainability, works council disclosure.

Published April 2, 2026

Article 13 of the EU AI Act requires high-risk AI systems to be sufficiently transparent for deployers to interpret outputs and use them appropriately. In HR and recruitment, transparency is fundamental to protecting candidates' rights and preventing discriminatory automation. Under Annex III, Section 4, AI systems used for recruitment, screening, filtering, or evaluating candidates, as well as systems making decisions on promotion, termination, or task allocation, are classified as high-risk and subject to full Article 13 transparency obligations.

This page explains how Article 13 transparency requirements apply specifically to HR technology — from CV screening to performance evaluation — including candidate notification, instructions for HR managers, works council disclosure, and what code-level violations Scanara detects in employment AI systems.

Which HR AI Systems Must Be Transparent?

CV Screening and Shortlisting

AI systems that filter, rank, or score candidate applications must explain which factors drove the ranking. “Opaque AI score: 72/100 — candidate rejected” without explanation of contributing criteria is non-compliant under Article 13(1).

Automated Interview Analysis

Video or voice analysis tools assessing candidates during interviews must disclose to candidates that AI is analyzing their responses, body language, or speech patterns. The analysis methodology and factors evaluated must be transparent.

Performance Evaluation Systems

AI systems monitoring employee productivity, generating performance scores, or recommending promotion and termination decisions must be transparent to both the employees being evaluated and the managers using the outputs.

Task Allocation and Workforce Management

Systems assigning tasks, shifts, or work distribution based on AI analysis must provide employees with information about how allocation decisions are made, particularly when affecting working conditions or compensation.

Employee Sentiment and Engagement AI

AI analysing employee communications or surveys to assess sentiment must inform employees of the monitoring and explain how the analysis is used. Note that emotion recognition in the workplace is prohibited under Article 5(1)(f) except for safety or medical purposes.

Key Transparency Obligations for HR Tech

Article 13 requires instructions for use that enable deployers to interpret AI outputs appropriately. For HR technology, this means instructions must be written for HR managers and recruiters — not just technical staff — and must explain how to correctly weight AI recommendations without succumbing to automation bias.

Candidates Must Know AI Was Used

Article 13(3) and Article 50(1) require that candidates are informed when AI is used in screening or shortlisting decisions. This complements GDPR Article 22, which gives individuals rights regarding automated decisions with legal or similarly significant effects. Notification must occur before or at the point of decision.

Scoring Factors Must Be Explainable

HR managers must be able to understand which factors drove an AI recommendation. Article 13(1) requires outputs to be interpretable — a candidate score without explanation of contributing criteria prevents the recruiter from making an informed, non-automated decision.

Works Councils Must Be Informed

In many EU member states, employee representatives must be informed before AI tools are deployed in the workplace. German BetrVG §87 and French loi travail require works council consultation on AI systems affecting working conditions. Article 13 instructions for use must be available to these bodies.

Instructions Must Prevent Automation Bias

Article 13 instructions must explain to HR managers how to correctly weight AI recommendations. Over-reliance on AI scores undermines human oversight required by Article 14. Instructions must specify when human judgment should override the system and how to identify potentially incorrect recommendations.

Regulatory Overlaps

HR AI transparency intersects with employment law, data protection, and anti-discrimination frameworks across EU member states. Article 13 compliance must account for these overlapping obligations.

GDPR Article 22 — Automated Decision Rights

GDPR gives individuals the right not to be subject to decisions based solely on automated processing with legal or similarly significant effects. Recruitment screening falls squarely within this scope. Article 13 adds system-level transparency requirements that complement GDPR's individual-level rights — both must be satisfied.

Equal Treatment Directives

The Racial Equality Directive (2000/43/EC) and Employment Equality Directive (2000/78/EC) prohibit discrimination in employment. Transparency about AI scoring factors is essential for demonstrating that the system does not discriminate on prohibited grounds. Opaque AI scores make it impossible to verify compliance with anti-discrimination law.

German BetrVG §87 — Works Council Co-Determination

German works constitution law grants works councils co-determination rights on technical devices monitoring employee behaviour and performance. AI recruitment and evaluation tools require works council agreement before deployment. Article 13 instructions for use must be provided to works councils to enable informed consent.

Platform Workers Directive (Proposed)

The EU Platform Workers Directive introduces transparency requirements for algorithmic management of platform workers, including explanations of automated decisions affecting working conditions, task allocation, and termination. These requirements reinforce and extend Article 13 obligations for gig economy AI systems.

Common Violations Scanara Detects

Candidate Rejection Without AI Disclosure

Rejection notification code that sends “Your application was unsuccessful” without disclosing that AI was involved in the screening decision. Article 13(3) and Article 50(1) require candidates to be informed of AI involvement — Scanara flags rejection endpoints missing disclosure fields.

Opaque Scoring Without Factor Explanation

API responses returning candidate scores without accompanying explanation of which evaluation criteria contributed to the score. Returning a numeric ranking without interpretable context prevents HR managers from using the output appropriately as required by Article 13(1).

Missing AI Interaction Disclosure in Interview Tools

Video interview platforms that analyse candidate responses without displaying a clear notification that AI is being used to evaluate the interaction. Scanara detects interview session initialization code that lacks a disclosure component or API header.

Instructions for Use Missing Limitation Documentation

System documentation that describes AI screening capabilities without documenting known limitations, such as accuracy degradation for non-standard CV formats, language biases, or career gap handling. Article 13(3)(b)(i) requires disclosure of limitations and foreseeable misuse scenarios.

Article 13 Compliance Checklist for HR Tech

Art. 13(1)

Make Outputs Interpretable for HR Managers

Ensure recruiters and HR managers can understand why the AI recommended or rejected a candidate. Provide factor-level explanations alongside scores, not just numeric rankings.

Art. 50(1)

Notify Candidates of AI Involvement

Inform candidates that AI is used in the screening or evaluation process. Notification should occur at the point of application or before AI analysis begins, in clear and accessible language.

Art. 13(3)(b)

Provide Instructions for Appropriate Use

Document how HR managers should weight AI recommendations, when human judgment should override the system, and how to identify potentially incorrect recommendations. Include known limitations and accuracy metrics.

BetrVG §87

Provide Transparency to Works Councils

Make Article 13 instructions for use available to works councils and employee representatives before deploying HR AI systems. Support co-determination processes with sufficient technical detail about system operation and impact.

Art. 13(3)(b)(ii)

Disclose Performance and Error Rates

Report accuracy, false positive/negative rates, and known biases in screening outcomes. Disaggregate by relevant demographic groups to demonstrate the system does not disproportionately disadvantage protected categories.

Frequently Asked Questions

How Scanara Helps

Scanara automates EU AI Act compliance from code to dossier. Connect your GitHub repos and get compliance reports in minutes.