EU delays AI Act employment-AI rules to December 2027

The European Parliament approved the provisional agreement on the Digital Omnibus on AI on 16 June 2026, pushing the application date for the AI Act’s “high-risk” requirements from 2 August 2026 to 2 December 2027. The high-risk classification covers AI systems used in recruitment screening, candidate ranking, promotion decisions, termination, task allocation, and performance monitoring. Council formal approval is still pending, with publication in the Official Journal expected before August 2026.

Why this matters

For European remote workers, freelancers and job seekers interacting with AI-driven recruitment platforms or workplace monitoring tools, the 16-month delay defers protections that were due to apply in six weeks. The notice requirements, the human-oversight obligations, and the non-discrimination safeguards built into the AI Act’s high-risk regime now arrive much later than originally legislated. Workers and applicants who would have had a right to know an AI system was being used to assess them will not have that right until late 2027 in most cases.

The delay is a deferral, not a repeal. The underlying obligations on AI providers and employers using these systems remain on the books, and the AI Act’s general-purpose AI rules and prohibited-AI rules already apply. National pay-transparency rules now in force across several EU member states (and applicable via direct effect against the State elsewhere) continue to govern algorithmic pay-setting where it intersects with disclosure obligations. Workers interacting with AI hiring systems before December 2027 should still be able to invoke GDPR Article 22 protections against fully-automated decisions with significant effects.

Employers running AI-assisted recruitment in Europe gain breathing room on compliance build-out, but the direction of travel is unchanged. For context on remote work and AI in Europe, see our AI remote marketing content guide and our hiring without an EOR explainer.