AI is now built into many of the tools hiring teams already use: applicant tracking systems, job boards, scheduling tools and video interview platforms. The useful question is which parts of a hire it should do, which it should help with, and which should stay with a person.
Since 2025 that has also become a legal question. This guide splits a hire into three kinds of work and sets out the EU and UK rules that apply to each, as they stand in October 2026. It is a summary, not legal advice.
Sort each task by who should decide
The line between the three kinds of work is about who carries the decision, not about what the technology can do. If a task has a clear right answer and a mistake is cheap to fix, automate it. If a tool can save time on a first draft but the result needs judgment, let it assist. If the outcome decides whether someone gets the job, keep a person responsible for it.
Automate
Tasks with a clear right answer, where a mistake is cheap to fix.
- Booking interviews and sending reminders
- Telling candidates where their application stands
- Collecting documents for pre-employment checks
Assist
The tool drafts or searches. A person checks and decides.
- Drafting job descriptions and adverts
- Summarising interview notes
- Suggesting people to approach
Keep human
Decisions about whether someone gets the job.
- Rejecting or shortlisting a candidate
- Assessing candidates in interviews
- Deciding and making the offer
The EU AI Act lists AI used to filter applications or evaluate candidates as high-risk (Annex III, point 4).
Most hiring processes have tasks in all three columns. The risk sits in the middle one, where a draft becomes the decision because nobody had time to check it.
List every step in your hiring process and put each one in a column before you buy a tool or switch on a new feature.
Automate the admin
Booking interviews, sending reminders, telling candidates where their application stands and collecting documents all take time, and all have a right answer. The interview is at ten on Tuesday or it isn’t. Automating them gives time back to the people running the process, and it helps candidates too: someone who hears nothing for three weeks usually assumes the answer is no.
On a senior search, diaries are often where the weeks go. Automated scheduling helps less when the panel is a CEO, a CFO and a board member, but it still removes the email chains around each slot. Our guide to where the time goes in a senior hire covers the rest.
Automate scheduling and status updates first, then apply to one of your own roles and read every message a candidate receives.
Let AI draft, and have a person decide
AI tools can produce useful first drafts: a job description from a brief, interview questions from a values sheet, a summary of an hour of interview notes, a list of people who might be worth approaching. Each one saves time, and each one needs someone who knows the role to check it.
Take more care when the draft is a judgment about a person. The EU AI Act lists AI systems intended for recruitment or selection as high-risk, in particular those used to place targeted job adverts, to analyse and filter applications, and to evaluate candidates. A tool that ranks your applicants or scores interview answers is likely to fall in that group.
From 2 December 2027, an employer using a high-risk system must:
- use it according to the provider’s instructions
- give human oversight to people with the competence, training and authority to do it
- monitor how it works, and keep its logs for at least six months
- tell workers’ representatives and affected workers before using it at work
That date was set by amendments made to the Act in 2026.
- 2 February 2025
EU: emotion recognition at work is banned
AI systems that infer the emotions of people in the workplace are prohibited, except for medical or safety reasons (AI Act Art 5(1)(f)).
- 5 February 2026
UK: new rules on automated decisions
A significant decision made with no meaningful human involvement needs safeguards, including the right to human intervention and to contest it (UK GDPR Arts 22A–22D).
- 2 December 2027
EU: high-risk duties apply to hiring tools
Employers using high-risk AI in hiring, such as tools that filter applications or evaluate candidates, must assign human oversight, keep logs and tell the people affected (AI Act Annex III and Art 26).
Dates as at October 2026. The EU date for high-risk systems was set by the 2026 Digital Omnibus amendments. The UK date is from the Data (Use and Access) Act 2025.
Source: AI Act Service Desk, Article 113
Ask each supplier whether their tool filters applications or evaluates candidates. If it does, plan for the EU duties now, while you can still choose how the tool is used.
Keep people on the decisions
Decisions about whether someone gets the job belong in the third column: who is rejected, who is shortlisted, who is offered the role and on what terms.
In the EU, one use is already ruled out. Since 2 February 2025, the AI Act has banned AI systems that infer the emotions of people in the workplace, except for medical or safety reasons. Don’t use any tool that claims to read how a candidate feels from their face or voice.
In the UK, the Data (Use and Access) Act 2025 rewrote the rules on automated decisions in the UK GDPR from 5 February 2026. A decision is based solely on automated processing when there is no meaningful human involvement, and it is significant when it has a legal or similarly significant effect on the person. An employer that makes a significant decision that way must give the person information about it, let them make representations, let them obtain human intervention and let them contest the decision. Stricter limits apply where the decision relies on special category data, such as health.
The test is whether the human involvement is meaningful. If the person who signs off sees only a score and approves every recommendation, it will be hard to show that it was.
Name the person who makes each rejection and offer decision, and make sure they see the evidence behind any score.
Tell candidates what you use
Both sets of rules expect you to tell people. Under the AI Act, an employer using a high-risk system to make or help make decisions about people must tell them they are subject to it. Under the UK rules, the first safeguard for a solely automated significant decision is information about it.
A clear explanation also does something the law doesn’t require. It tells the candidates you most want that a person will read what they send.
Add two sentences to your candidate privacy notice: which AI tools you use and for what, and how a candidate can ask for a person to review a decision.
Decide the columns before you buy the tools
A senior hire has fewer tasks to automate than a volume hire, and more that depend on judgment: who to approach, how to read a career, whether this person suits this business now. The admin can still be automated. The judgment can’t be handed over.
On our Head of HR searches, Matteo approaches each candidate, explains why they are worth meeting and stays involved through interviews, references and the offer. Our Head of HR search page explains how that works.
Sources
- Regulation (EU) 2024/1689 (Artificial Intelligence Act), EUR-Lex: https://eur-lex.europa.eu/eli/reg/2024/1689/oj
- European Commission, AI Act Service Desk, “Article 5: Prohibited AI practices”: https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-5
- European Commission, AI Act Service Desk, “Article 26: Obligations of deployers of high-risk AI systems”: https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-26
- European Commission, AI Act Service Desk, “Article 113: Entry into force and application”, as amended by the Digital Omnibus on AI: https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-113
- European Commission, AI Act Service Desk, “Annex III: High-risk AI systems referred to in Article 6(2)”: https://ai-act-service-desk.ec.europa.eu/en/ai-act/annex-3
- Data (Use and Access) Act 2025, section 80, legislation.gov.uk: https://www.legislation.gov.uk/ukpga/2025/18/section/80