Skip to main content
Employment & HR · Annex III §4

AI Act readiness for hiring teams and recruiters.

If your firm uses AI to screen CVs, rank candidates, target job ads, or evaluate staff, you're operating under Annex III §4 of the EU AI Act, and some hiring tools are prohibited outright. cosantio gives HR teams, internal recruiters, and agencies the workflow, artefacts, and named officer to be audit-defensible in time. Examples here are drawn from the Irish market we know best, but the obligations apply across the EU, with some country-specific variation.

What governs AI in employment & HR

Four pieces of regulation. One workflow.

Hiring AI in Ireland and the EU is regulated by more than the AI Act. The four frameworks below overlap, and one of them bans certain tools outright. cosantio's Scorecard maps your firm against all four at once.

01From 2027

EU AI Act, Annex III §4

Source: Regulation (EU) 2024/1689

Classifies AI used for recruitment, selection, and workforce management as high-risk: targeting job ads, screening and ranking applicants, and decisions on promotion, termination, task allocation, or performance monitoring. Triggers obligations on data governance, human oversight, transparency, and record-keeping.

High-risk regime applies02 Dec 2027
02In force

Article 5 prohibited practices

Source: AI Act Art. 5 · since Feb 2025

Some hiring tools are banned, not merely high-risk. Inferring a candidate's emotions in an interview is prohibited under Article 5(1)(f), and the Commission's guidelines extend “workplace” to recruitment. This applies now and is not deferred to 2027.

StatusAlready binding
03In force

GDPR & automated decisions

Source: GDPR Art. 22 · DPC

Candidate and employee data is personal data. Article 22 restricts solely-automated decisions with legal or significant effect, and a DPIA is usually required. The Data Protection Commission supervises hiring analytics, profiling, and employee monitoring.

Supervised byDPC
04In force

Employment equality law

Source: Employment Equality Acts

A screening or ranking tool that produces discriminatory outcomes across gender, age, disability, race, or other protected grounds breaches equality law regardless of the AI Act. Enforced in Ireland by the WRC and IHREC, with the burden often on the employer to justify.

Enforced byWRC · IHREC
Who's in scope

Hiring teams, agencies, and vendors. Different obligations.

The AI Act treats your firm differently based on whether you develop the AI or use it. That distinction (provider vs deployer) shapes which Article 16 or Article 26 obligations apply, and building or fine-tuning a screening tool can pull you into both.

Typically Deployer

Internal hiring teams.

HR and hiring managers in SMEs using a third-party ATS, CV-screening, or scheduling tool. Acting as deployer under Article 26 means human oversight, transparency, logging, and the worker-information duty, even though you did not build the tool.

Key obligation
Human oversight and worker notice. Inform workers and their representatives before use, under Article 26(7).
Deployer, often Provider

Recruitment agencies.

Agencies screening, matching, or ranking candidates on behalf of clients. Build a screening model, or fine-tune a vendor's on your own candidate pool, and Article 25 can reclassify you as a provider with the full Article 16 obligation set.

Key obligation
Resolve the vendor-versus-provider question first, then classification, bias testing, and candidate transparency per role.
Provider

HR-tech vendors.

Firms that build the screening, matching, or performance-management AI that everyone else deploys. Full provider obligations apply: technical documentation, data governance, conformity assessment, and EU registration.

Key obligation
Annex IV technical dossier per system, plus the instructions and documentation your deployers need to comply.

Worker-notice duty. Before using a high-risk AI system at work, Article 26(7) requires employers to inform workers' representatives and the affected workers first. This is a deployer obligation that lands even when the tool is a third-party product you did not build. The Scorecard flags where it applies to your firm.

In-scope AI systems

The systems hiring teams actually run.

Real systems we see across Irish and EU employers and agencies. Classifications and obligations below are illustrative; your firm's classification depends on the specific use, not the product name.

CV screening & candidate rankinge.g. ATS shortlisting, scoring
AI that filters applications, scores CVs, or ranks a shortlist for human review. The headline §4(a) use case.
High-risk
Targeted job advertisinge.g. audience-targeted ad delivery
AI that decides which candidates see a job advert. Named explicitly in §4(a), and a route to unlawful indirect discrimination if the audience is skewed.
High-risk
Performance & workforce managemente.g. monitoring, task allocation, evaluation
AI that monitors, evaluates, allocates tasks, or informs promotion and termination decisions. The §4(b) workforce-management use case.
High-risk
Emotion & sentiment analysise.g. video-interview “cultural fit” scoring
AI that infers a candidate's emotions, confidence, or engagement from face, voice, or body language is prohibited in a hiring context, not merely high-risk.
Prohibited
HR chat & schedulinge.g. candidate FAQ bots, interview booking
Candidate-facing AI for FAQs, scheduling, and routing. Not a hiring decision, but an interaction that must be disclosed.
Limited-risk
Sample hiring dossier

What an audit-defensible hiring file looks like.

Excerpt from a sample Annex IV technical dossier for a candidate-screening engine. Every finding cites firm-specific evidence; every gap names an owner and a date.

Technical Report · Annex IV (EU) 2024/1689

EU AI Act Conformity · Candidate Screening

Provider: Sample Recruitment LtdSector: Employment & HR · Recruitment agency
CSC-2026-0419Issued 13 May 2026
System
TalentRank (v3.2)
Intended purpose
CV screening and candidate ranking for SME hiring
Hardware
AWS EC2 · EU-west-1
Oversight
Recruiter in the loop · mandatory human review before any rejection
01Risk ManagementArt. 9 · 3 questions
Is there a documented risk management system covering the full lifecycle?
Compliant
Cited evidence
Risk Management Framework v2.4 (Q1 2026) covers design, deployment, and post-market phases. Reviewed quarterly by the Risk Committee, chaired by the COO.
Is there a process for post-market risk monitoring?
Action required
Remediation plan
By end of Q3 2026, stand up a monthly fairness and drift monitor on shortlist outcomes with alerting to the Compliance Officer. Owner: A. Murphy (Head of People). Reviewed against the Employment Equality Acts.
03Data & Data GovernanceArt. 10 · 4 questions
Are training and input datasets tested for bias across protected groups?
Compliant
Cited evidence
Quarterly disparate-impact testing on shortlist rates by gender, age, disability, and ethnicity. Reviewed by the People & Risk Committee; last run Q1 2026 with no material disparity flagged.
Is data provenance documented for every feature in the model?
Action required
Remediation plan
By end of Q2 2026, complete the feature-lineage register for all model inputs, removing any proxies for protected characteristics and flagging social-media-derived features for an Article 5 social-scoring review. Owner: D. Byrne (Data Lead).
+ 9 more pillars in the full platform dossier · 25 questions across Annex IV
What you'll get

Start with the Scorecard. Go deeper in a demo.

The Scorecard is the same eight questions for every sector, but the result is sector-specific. Hiring teams get §4 classification, a prohibited-practice check, and the Article 26(7) worker-notice obligations layered on top of the core readiness assessment.

Recommended start

Fast Scorecard.

Free · 3 minutes · no login

Eight plain-English questions. A regulator-grade PDF that scores your firm against every AI Act obligation in scope, with hiring-specific framing throughout.

  • Classification under §4, with a prohibited-practice flag
  • The Article 25 vendor-versus-provider check
  • Status on Article 4, 5, 50, and 26(7) obligations
  • Prioritised action plan with weeks-to-readiness estimate
~3 min · PDF in 5Start the Scorecard
Deep dossier

Deep Dossier.

Demo · 20 minutes · with our team

The full Annex IV workflow inside the platform: 25 questions across 11 pillars per AI system, producing a regulator-grade technical dossier for an auditor, a client, or your board. Book a demo to walk through a sample and the hiring-specific features.

  • Annex IV-shaped technical dossier per system
  • Cited evidence and remediation plans per finding
  • Sample candidate-screening pillar map (TalentRank reference)
  • Signed cover sheet for board and regulator use
~20 min · live walkthroughBook a demo
Common questions

Things hiring teams ask us first.

Employment & HR · §4

Find out where your hiring stands.

Eight questions. Three minutes. A regulator-grade PDF with §4 classification, a prohibited-practice check, and a prioritised action plan for your firm.