Career Ahead Magazine
Subscribe
Government & Policy

The AI Hiring Rules Europe Promised Are Now Due in 2027

Regulation EU 2026/1744 pushed AI Act high-risk employment obligations from August 2026 to December 2027. What the delay means for job applicants in Europe.

The flag of the European Union
Photo: Pexels

Brussels moved the deadline. Regulation (EU) 2026/1744, signed on July 8, 2026, pushed the application date for the AI Act’s high-risk employment obligations from August 2, 2026 to December 2, 2027. The original date has already passed. Every vendor that told corporate customers its hiring software would be compliant by last summer is now operating on a different calendar, and almost no one told the job applicants.

Flag of the European Union
Photo: Pexels

What got postponed

The obligations affected are in Sections 1, 2 and 3 of Chapter III of the AI Act. They cover AI systems classified as high-risk under Article 6(2) and listed in Annex III. Point 4 of Annex III names, verbatim, systems intended for “the recruitment or selection of natural persons, in particular to place targeted job advertisements, to analyse and filter job applications, and to evaluate candidates.” It also covers systems used “to make decisions affecting terms of work-related relationships, the promotion or termination of work-related contractual relationships, to allocate tasks based on individual behaviour or personal traits or characteristics or to monitor and evaluate the performance and behaviour of persons in such relationships.”

Those systems now have a later compliance date. For high-risk systems under Article 6(2) and Annex III, the obligations apply from December 2, 2027. For high-risk systems used as safety components in products covered by EU sectoral legislation, under Article 6(1), they apply from August 2, 2028. The change came through what law firms call the AI Digital Omnibus, a set of amendments the Council and Parliament agreed in June 2026 and which entered into force in August 2026.

What is already live

The delay does not touch the AI Act’s bans. Under Article 5(1)(f), AI systems that infer emotions in the workplace and in education have been banned in the EU since February 2, 2025, the first application date set by the Act’s phased timetable. The omnibus did not move that.

Other transparency obligations are also in force. Since August 2, 2026, providers of AI systems placed on the EU market must disclose when individuals are interacting with an AI system and mark deepfakes, under Article 50. Watermarking obligations for AI-generated content apply from December 2, 2026. What remains postponed is the heavy regulatory machinery on the high-risk employment systems themselves: risk management, data governance, technical documentation, conformity assessment, registration in the EU database, and market surveillance.

Why Brussels delayed it

The Commission’s stated reason is candid. The recital to Regulation 2026/1744 says the delayed availability of standards, common specifications, and alternative guidance, together with the delayed establishment of national competent authorities, jeopardizes effective application and risks a significant increase in implementation costs. In other words, Europe wrote the most ambitious AI employment law in the world, then postponed it because the machinery to enforce it did not exist.

That framing deserves scrutiny. Law firms tracking the transition describe the delay as a response to real implementation gaps, and that is plausible. But the same delay means the world’s first comprehensive AI employment law is being enforced on a rolling basis that no one outside legal practice has fully mapped. The credibility of the AI Act as a regulatory model depends on whether December 2, 2027 holds.

A European city
Photo: Pexels

What the delay means for applicants

For anyone applying for a job in Europe, the consequence is concrete. The software reading a CV, scoring a video interview, or ranking an application is not yet governed by the AI Act’s employment provisions. An employer in Berlin, Madrid or Dublin can still run a screening system in 2026 and 2027 without the technical documentation, bias testing, and market surveillance the Act will eventually require. The ban on workplace emotion inference is law; the rules on how hiring algorithms must be built and audited are not.

The consequences run both ways. Employers buying screening tools now have sixteen extra months before the high-risk obligations apply, which may soften their negotiating position with vendors. Vendors, in turn, can continue selling systems without the technical documentation and post-market monitoring the Act will eventually require. Neither side has an incentive to publicize the gap.

For workers aged 16 to 35, the practical result is uncertainty. A 22-year-old submitting applications in 2026 has fewer statutory protections than the law’s architects originally intended. The EU classifies employment-related AI among high-risk systems, but high-risk on paper is not the same as high-risk in force. The Act also reaches beyond the EU, because its rules bind any provider placing systems on the EU market, including the global HR technology vendors whose software is used to hire people everywhere.

The counterargument

The delay is not purely a loss for workers. Standards bodies were late; national regulators were not ready; and applying the full regime before the support structure existed could have produced inconsistent enforcement across member states, which is arguably worse for everyone. Some legal commentators also note that the extra period gives vendors and employers time to design systems that meet the rules instead of scrambling. The open question is whether that time will be used for compliance or for holding the line.

What to watch

Whether member states designate their market surveillance authorities before the new date arrives. The Netherlands has proposed a Platform Work Act; Ireland completed a public consultation on transposition of the Platform Work Directive. The first enforcement action against a hiring vendor under Chapter III will test whether the delayed timetable was prudent or simply postponed accountability. Until then, the most useful thing an applicant can do is ask what software evaluates their application, and what bias testing that vendor has published.

This article was prepared with AI assistance and edited and fact-checked by Poornima Singh, Career Ahead.

Share this story
Career Ahead Editorial Team

Written by

Newsroom, Career Ahead Magazine

The Career Ahead Editorial Team is the newsroom of Career Ahead Magazine, a career and leadership title in print and online since July 2020. Our editors report and edit the interviews and career guides that first-person contributors don't write. You'll find us writing for students choosing what to study and for professionals planning their next move. We also cover founders…

Career Ahead follows strict sourcing and fact-checking standards. About our editorial standards.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

→

More in Government & Policy

View all →

Connect with us

Say hello →

Be Ahead

Sign up for our newsletter

Get regular updates directly in your inbox!

We don’t spam! Read our privacy policy for more info.