Guide
Using AI in Recruitment: A UK Employer's Checklist
You can use AI in hiring, but you stay responsible for every decision it touches. Before you switch a tool on, check five things. What data does it use? Does a person review outcomes? How do you test for bias? What do you tell candidates? How can they challenge a result?
Why is AI in recruitment risky?
Hiring decisions affect people's livelihoods, and the law treats them seriously. An AI tool can repeat past bias, reject good candidates for odd reasons, or make a decision no one can explain.
In March 2026 the ICO published a report on automated decision making in hiring. It expects organisations to be open about their use of AI, to test for bias and to let candidates challenge decisions. You can read the ICO's statement.
What should I check before I use an AI hiring tool?
Work through this list:
- 1
Know what the tool does
Does it only sort and schedule, or does it score or reject people?
- 2
Keep a person involved
A person should review any decision that affects a candidate. The ICO defines automated decision making as a decision made with no human involvement.
- 3
Ask the supplier about bias testing
The ICO suggests asking developers how they test before you buy.
- 4
Do a data protection impact assessment
Hiring tools process personal data at scale. A DPIA is expected where processing is likely to be high risk. See the ICO's DPIA guidance.
- 5
Tell candidates
Say in your privacy notice and job advert when AI is used, and what for.
- 6
Give a route to challenge
Tell candidates how to ask for a human review.
- 7
Test regularly
Check outputs for bias on a schedule, not once.
- 8
Keep records
Note what the tool did, who reviewed it and why you decided.
What does the law say about bias?
The Equality Act 2010 protects people from discrimination based on protected characteristics. This applies to AI decisions as it does to human ones. A tool that disadvantages one group can put you in breach, even if you did not intend it. See the government's Equality Act guidance.
Rules on solely automated decisions in UK data protection law were changed by the Data (Use and Access) Act 2025. The ICO has been updating its guidance. Check the current version before you rely on a tool that decides alone.
Where can AI help safely?
Low risk jobs are a good start:
Booking interviews and sending reminders
Parsing CVs into your applicant system, with a person reviewing the result
Drafting job adverts for a person to edit
Right to work and onboarding paperwork
Keep people in charge of shortlisting, rejection and offers.
Common questions
Can AI reject candidates on its own?
It is risky. If a tool decides with no human involvement, you need safeguards and a clear legal basis. Keep a person in the loop.
Do I have to tell candidates I use AI?
Yes, you should be open about it. The ICO expects transparency.
Do small firms need a DPIA?
If processing is likely to be high risk, yes, whatever your size. Hiring tools often fall in that group.
Does the EU AI Act affect me?
It may if you recruit in the EU. Recruitment tools are treated as high risk there. See our EU AI Act guide.
Is this legal advice?
No. It is a checklist. Ask a qualified adviser about your own case.
How HAeX can help
We build hiring workflows with human sign off built in, and we help write the policy. See AI for recruitment and HR, or draft a policy with our free AI Policy Builder.
If you need a specialist to build or run a hiring tool, see AI recruitment.
Talk through AI in your hiring
The first call is free. We tell you what to check before a tool screens candidates.