Issue 14 · Nov 2–8, 2026Agent Age News magazine · Other editions · Search · Method · HU / ENG

Work changes twice in the age of AI: first what people do, then what they are paid for.

regulation · European Union

EU AI Act: recruitment AI is high-risk, but some obligations get a delay

Sep 21, 2026 · Issue 8 · 7 sources listed · 10 articles in the group

The EU AI Act has applied generally since 2 August 2026, but obligations for high-risk employment systems have been pushed to a later date in 2027, a reprieve of about 15 months. The transparency obligation is already in force: candidates must be told if they are dealing with an AI system, unless this is obvious.

Who is affected and how

job seeker
You have the right to know if an AI assistant is screening your application.
manager
Compliance documentation is an important criterion when choosing recruitment AI vendors.
HR professional
Start now on AI notices for applicants and on assessing high-risk systems.
SME owner
The delay is worth using to clarify vendor compliance.

Analyst's note

If the output of recruitment AI materially influences hiring decisions, the system is considered high-risk, even if a human makes the final decision.

The full analyst's noteannouncement claim examined: correlation

What it rests on

Recruitment AI systems automatically fall into the high-risk category if used for screening, ranking, or evaluating candidates. The regulation covers any tool that materially influences employment decisions, even if the final decision is made by a human.

What would refute it

It would refute the claim if a recruitment system that filters and ranks candidates were not classified as high-risk in a regulatory review because its output does not materially influence the decision. It would also refute the claim if the regulation excluded from the high-risk category systems that perform only a preparatory or supporting function and do not carry out profiling.

What to watch

The situation will be clarified by how recruitment organizations document human oversight mechanisms in candidate evaluation. Compliance with regulations will be indicated if companies can demonstrate the extent to which the system's output influences the final decision and what steps they take to correct potential biases.

Perspective

The interests of companies involved in recruitment and candidates are in the foreground, especially legal compliance and transparency. According to the sources, companies must assess risks, and candidates have the right to know if their applications are screened by an AI system. Internal business strategies of recruitment software vendors and details of competitors' market battles are omitted.

Rival hypotheses

The 15-month delay could also favor the consolidation of the recruitment AI market and the strengthening of major providers, as indicated by acquisition activity. The immediate entry into force of transparency obligations could also act as a market constraint, as the lack of candidate notification already creates legal risk before the high-risk classification takes effect.

Red team

Edited counter-argument
EdgeTal's independently unverified claims about its own architecture, regarding on-device processing and a private LLM, appear in the article as evidence supporting compliance, yet the source base consists exclusively of the provider's marketing material. The description of the exceptions under Article 6(3) is fragmentary in the source text, and the concept of narrow procedural, preparatory or support functions remains undefined, making the scope of the high-risk classification difficult to assess.

Vakfolt

Specific risks for small businesses and startups are omitted, as their resources for compliance may be scarce. Sources do not address how using recruitment AI systems affects employees' mental health or workplace culture during long-term adaptation.

Source profile

10 articles · 8 publishers · 7 independent pieces (8 press); of these, Hungarian: 0. Of the 10 articles, 8 links are shown. Published: 2026-09-21 – 2026-09-25.

We found no primary document (an official, court or scientific publication, or a company statement) among the sources.

Measured blind spot

  • There is no Hungarian-language or Hungarian source among the listed sources.
  • All listed sources were published in one language.

Background

Machine analytical reading based on the story summary and the linked sources, following editorial rules. It contains no new facts; the facts come from the linked article. How is it made?

Narrative: The employer is responsible for algorithmic decisions

Sources

Machine-generated summary based on the titles and excerpts of the source articles. The facts come from the linked article.

What do you make of this piece?

The split shows after 20 votes.

Comment on LinkedIn
Saved pieces

    Weekly email

    One email a week with the main stories from the Agent Age editions. You confirm by email and can unsubscribe in one click.

    Almost done: we sent you an email. Open it and press the confirm link. If it does not arrive within a few minutes, check your spam folder.