AI in Recruitment

Your next job rejection might be illegal — and you’d never know it, because the decision wasn’t made by a person.

Across the UK, AI-powered candidate screening tools are rejecting thousands of job applications every day. Employers trust them because they’re fast, scalable, and carry the appearance of objectivity. But objectivity isn’t the same as legality — and when these algorithms are tested against Equality Act standards, the results are alarming.

This guide explains what AI recruitment tools actually do, where they fail UK law, and exactly how candidates and HR professionals can challenge automated hiring decisions.

The Myth of Algorithmic Neutrality

AI screening tools are trained on historical hiring data. The problem is that historical hiring data reflects historical human bias — a bias that the algorithm doesn’t question, it simply replicates at scale.

A system trained on a decade of a company’s past hires will learn to favour candidates who look like the people who were hired before. If the company historically hired fewer women into technical roles, more candidates from Russell Group universities, or fewer people with career gaps, the algorithm encodes those patterns as signals of quality. It isn’t programmed to discriminate — it learns to.

The result is indirect discrimination delivered at machine speed, with no human in the room to question it.

What the Audit Reveals

When identical CVs are submitted to AI screening tools with altered demographic indicators — a name that reads as female rather than male, an address in a lower-income postcode, a career gap that corresponds to a period of maternity leave — the match scores change. Sometimes dramatically.

The same qualifications. The same experience. The same skills. Different scores, based on data points the algorithm was never supposed to weight.

The parameters that most commonly trigger discriminatory outcomes include:

Career gaps — which disproportionately impact women (maternity leave), disabled candidates (health-related absence), and older workers

Non-linear education pathways — which disadvantage candidates from lower socioeconomic backgrounds

Address and postcode data — a proxy for socioeconomic status that has no legitimate place in skills-based screening

Employment history patterns — which penalise sectors and roles more commonly held by protected groups

How This Maps to the Equality Act

Under the Equality Act 2010, indirect discrimination occurs when a provision, criterion, or practice that appears neutral puts people with a protected characteristic at a particular disadvantage — and cannot be objectively justified.

An AI screening algorithm is a provision, criterion, or practice. If it systematically disadvantages candidates on the basis of characteristics connected to a protected group, it is indirect discrimination — regardless of whether a human intended it.

Critically, employers cannot outsource their legal liability to a software vendor. Using a third-party AI tool does not transfer the duty of compliance. The employer remains responsible for every hiring decision the tool produces. If the tool discriminates, the employer discriminates.

Recent regulatory developments under UK GDPR also give candidates meaningful rights. Under Article 22 of UK GDPR, individuals have the right not to be subject to a decision based solely on automated processing that significantly affects them — including job rejections. Candidates can request human review of any automated rejection, and employers must comply.

How Candidates Can Challenge an Automated Rejection

If you believe an AI tool may have rejected your application unfairly, here are the steps to take:

Submit a Subject Access Request (SAR) under UK GDPR — request all personal data the employer holds about you, including any automated decision-making data and the logic involved

Request human review — under Article 22 UK GDPR, state clearly that you are contesting the automated decision and requesting a human to review your application

Ask for the criteria — request the specific criteria and weighting used in the screening process

Document your protected characteristic — if the disadvantage connects to a protected characteristic under the Equality Act, set this out clearly in writing

Contact ACAS — if the employer fails to respond or refuses to provide the information, ACAS Early Conciliation is the first step before an employment tribunal claim

What HR Professionals Must Do

If your business uses AI screening tools, your legal exposure is real and growing. Steps to take now:

Audit your screening tool against Equality Act standards — request the vendor’s bias testing documentation

Ensure human review is available to any candidate who requests it

Remove demographic proxies from your screening criteria — postcode, career gap penalties, education routing

Document your justification for every criterion the tool applies

Review your data retention and SAR response processes — a 30-day response window is mandatory