EU AI Act for Recruiting: AI Screening and Hiring Calls

Is AI in recruiting high-risk under the EU AI Act?

Much of it is. Annex III point 4(a) lists "AI systems intended to be used 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". Few areas in the Act are named this directly.

But not every AI that touches a candidate is a selection system. An agent that calls back an applicant to find an interview slot is doing admin, not selection. The Act gives a way to treat that kind of system as not high-risk, with conditions. The question for every recruiter is which side of that line each part of the process sits on.

Where is the line between scheduling and screening?

The "can be outside" rows rest on Article 6(3). An Annex III system is not high-risk when it poses no significant risk, including by not materially influencing the outcome of decisions, and it does one of four things: a narrow procedural task, improving the result of a finished human task, spotting patterns without replacing human judgment, or a preparatory task. Under Article 6(4) the provider who relies on this must document the assessment before the system goes to market, is subject to registration, and must show the assessment to authorities on request.

Why does profiling matter so much in hiring?

Because it closes the exit. Article 6(3) ends with this sentence: an AI system referred to in Annex III "shall always be considered to be high-risk where the AI system performs profiling of natural persons." Profiling means automated processing of personal data to evaluate or predict aspects of a person, such as work performance, reliability or behavior.

That is what most screening does. A voice agent that asks about experience and gives a fit score, sorts candidates into "strong" and "weak", or decides who gets a second interview is evaluating people. So for recruiting, the practical test is simple: if the AI's output says anything about how good a candidate is, treat it as high-risk.

Can an AI screening call read a candidate's emotions?

No. Article 5(1)(f) bans AI that infers people's emotions in the area of the workplace, except for medical or safety reasons, and it has applied since 2 February 2025. The European Commission's guidelines on prohibited practices treat recruitment as part of the workplace for this ban, as summarized by the Future of Privacy Forum.

That rules out features that claim to read a candidate's enthusiasm, nervousness or stress from their voice. A banned practice can be fined up to EUR 35 million or 7% of worldwide annual turnover, whichever is higher (Article 99(3)).

What must a recruiting AI agent tell candidates?

Every AI agent that talks to people must be designed so they know they are dealing with an AI, unless that is obvious (Article 50(1)), and the information must be clear and given at the latest at the first interaction (Article 50(5)). This applies from 2 August 2026, whatever the risk level.

A clear opening for a scheduling call: "Hi, this is the AI assistant of [Agency], calling about your application for the warehouse role. I can book your interview now, or a recruiter can call you back." If the system is high-risk and helps make decisions about the candidate, Article 26(11) adds a second duty from December 2027: tell candidates that a high-risk AI system is used on them.

If your recruiting AI is high-risk, what must you do?

An agency or employer that uses a high-risk system built by a vendor is its deployer. These duties apply from 2 December 2027.

GDPR applies alongside all of this, including its rules on decisions based solely on automated processing. Breaching the deployer duties or the disclosure duty can be fined up to EUR 15 million or 3% of worldwide annual turnover; for SMEs and start-ups the lower amount applies (Article 99).

What are the key dates for recruiters?

The high-risk date was 2 August 2026 in the original text. Regulation (EU) 2026/1744 moved it to 2 December 2027. The ban and the disclosure duty did not move.

How should a recruiting AI phone agent be set up?

Agencies use phone agents for speed: calling back applicants within minutes, filling interview slots, reminding people the day before. All of that can be done without the agent ever judging a candidate.

• Let the AI schedule and inform. Booking, moving and confirming interviews, answering questions about the role, pay range you published, location and next steps.
• Keep the decision with the recruiter. No scores, rankings or "fit" labels from the phone agent.
• No automatic knockouts unless your counsel has signed off and you run that flow as high-risk. The agent can record answers to required questions; a recruiter reviews them.
• No voice-based emotion features, at any stage.
• Disclose in the first sentence, and offer a callback from a recruiter.
• Write down the classification of each flow, and update it when the flow changes.

How does an AI phone agent from AInora fit?

AInora builds AI phone agents for staffing agencies and HR teams: calling applicants back, booking and confirming interviews, reminders and answering questions about the process. How it lines up with the rules above:

• The agent says it is an AI assistant in its first sentence, in the words you choose.
• It hands the call to a recruiter when the candidate asks, or when your rules say so.
• Every call gets a transcript and a summary your recruiters can review.
• You decide what it may and may not ask. If scoring is off the table, it stays off the table.
• It speaks many languages, which helps with international candidate pools.

Your counsel decides the classification; the agent follows the rules you set. To hear one on your own roles, fill out the short form and we build a demo agent from your website. More on the use cases: AI voice agents for staffing and recruiting.