Why does the EU AI Act treat hiring AI as high-risk?

The Act puts AI used for recruitment, selection, promotion, termination and task allocation into its high-risk category because these systems affect people's livelihoods and are prone to bias that can disadvantage protected groups. High-risk status brings real obligations rather than a light touch.

Does the EU AI Act apply to UK recruiters?

Sometimes directly, where your use has an EU dimension such as placing candidates into EU roles. For purely UK hiring it may not apply directly, but the tools you buy are being built to the Act and the UK has overlapping rules, so it is wise to treat its expectations as the standard.

What does the EU AI Act expect of recruiters using AI?

Human oversight so a person makes the actual decision, transparency so candidates can learn AI is used, records of how the system is used, accuracy and bias monitoring of real outcomes, and using tools as the provider intended. UK GDPR and the Equality Act 2010 already require much of this.

Resources · 26 June 2026

The EU AI Act and AI hiring tools: what UK recruiters need to know

By the Operating Bench Team · Last reviewed 26 June 2026

The EU AI Act singles out AI used in recruitment and employment as high-risk. If you place candidates, screen CVs with a tool, or rank applicants with software, this is worth ten minutes of your attention, because even as a UK recruiter it can reach you, and it sits on top of the UK rules you already have to follow.

Why hiring AI is “high-risk”

The Act puts AI used for recruitment and selection, and for decisions about promotion, termination and task allocation, into its high-risk category. The reasoning is simple: these systems affect people’s livelihoods, and they are prone to bias that can quietly disadvantage protected groups. High-risk status brings real obligations rather than a light touch.

Does it apply to UK recruiters?

The honest answer is: sometimes directly, and increasingly as the benchmark either way.

It can bind you directly where your use has an EU dimension, for example when you place candidates into roles in the EU, or use a system whose output is relied on there.
For purely UK hiring it may not apply directly, but two things make it matter anyway: the vendors building these tools are designing to the Act, so its requirements arrive in the products you buy; and the UK has its own overlapping rules that point in the same direction.

The practical stance is to treat the Act’s expectations as the standard, because you will likely be held to something very like them regardless.

What it expects of you (the deployer)

Human oversight. A person makes the actual decision, with the tool as input.
Transparency. Candidates should be able to learn that AI is used and what it means for them.
Records. Keep logs of how the system is used.
Accuracy and bias monitoring. Watch real outcomes and act if a tool disadvantages a group.
Use it as intended. Follow the provider’s instructions rather than improvising.

The UK rules that already apply

You are not waiting for anything to start here:

UK GDPR restricts decisions based solely on automated processing where they have a significant effect, such as rejecting a candidate, so keep a meaningful human in the loop.
The Equality Act 2010 makes you responsible if a screening tool causes indirect discrimination, and “the algorithm did it” is not a defence.

We cover that ground in detail in Is AI candidate screening legal in the UK?

What to do

List the hiring AI you use, including features built into your ATS.
Keep a human decision-maker on every shortlist and rejection.
Tell candidates AI is used, in plain terms.
Check your vendors for oversight and bias features, and for proper data terms.
Monitor outcomes for adverse impact.
Write it into your AI policy so it is consistent and you can show your reasoning.

That last step is where a policy earns its keep. Our AI Safe-Use Pack for recruitment agencies includes a risk register that already names screening bias and automated-decision risk, plus the policy and staff guidance to support it. For the essentials first, the general AI Safe-Use Pack covers the core, and the free AI Starter Kit gives you the ground rules on one page.

This article is general information, not legal advice. How the EU AI Act, UK GDPR and the Equality Act apply to your agency should be considered with professional advice.