EUAIACTUS.COM

Annex III • employment

EU AI Act compliance for HR, recruitment and workforce AI

Employment is a core Annex III area. AI used for recruitment, candidate evaluation, employment decisions, task allocation or worker monitoring can fall within the high-risk framework depending on intended purpose and the Article 6 analysis.

Updated and legally reviewed: 26 September 2026.

Direct answer

HR AI can require both EU AI Act high-risk analysis and separate GDPR review.

Recruitment, candidate evaluation, promotion, termination, task allocation and worker monitoring are central Annex III employment use cases. Where the same systems process personal data, profiling, Article 22, DPIA, transparency and lawful-processing questions may arise separately under GDPR.

Annex III employment uses

The AI Act specifically lists several HR use cases.

Recruitment and selection

AI intended for recruitment or selection, including targeted job advertising, analysing/filtering applications and evaluating candidates, is within the Annex III employment category.

Employment decisions

AI intended for decisions affecting terms of work, promotion or termination can fall within the high-risk listing.

Task allocation

AI allocating tasks based on individual behaviour or personal traits/characteristics is specifically relevant to the Annex III employment category.

Monitoring and evaluation

Systems intended to monitor or evaluate performance and behaviour of persons in employment relationships can fall within the listed use case.

Classification nuance

Not every HR tool is automatically high-risk.

Classification depends on whether the system falls within the listed use case and on the Article 6 filter/qualifications. Providers relying on a non-high-risk conclusion for an Annex III-listed system should document the assessment. Profiling of natural persons receives specific treatment under Article 6.

Current timeline

The main Annex III high-risk regime applies from 2 December 2027.

Regulation (EU) 2026/1744 moved the employment/high-risk application timetable. But organisations using HR AI should not wait: Article 50, AI literacy, GDPR/employment law and other currently applicable obligations can already matter.

Readiness controls

What employers and HRTech providers should build now

Use-case inventory

Map each HR AI system to recruitment, screening, scoring, monitoring, promotion, termination, task allocation and other intended purposes.

Role mapping

Separate the HRTech provider's duties from the employer/deployer's duties and document upstream/downstream dependencies.

Data/bias testing

Assess data quality, performance, discriminatory impacts, error patterns and subgroup outcomes appropriate to the system/use context.

Human oversight

Define when and how human reviewers can understand, challenge, override or stop AI-supported employment decisions.

Candidate/worker transparency

Map AI Act, GDPR, employment-law and local notice obligations rather than relying on a single generic AI disclosure.

Vendor evidence

Require documentation, change notices, performance information and compliance cooperation sufficient for your own deployer obligations.

Cross-border HR

U.S. employers can face overlapping regimes.

An organisation hiring in Europe can face EU AI Act and GDPR obligations; a New York City use case can separately raise NYC Local Law 144 questions. Do not assume one audit or notice satisfies every regime.