Hiring is one of the most time-consuming jobs in a small business. A single job advert can bring hundreds of applications, and reading them all properly takes days. AI tools can now read CVs, summarise experience, match candidates to requirements and schedule interviews in minutes.

Used carelessly, they can also reject good candidates unfairly and break the law. Here's how to use AI in recruiting the right way. This is general information; for your situation, take HR or legal advice.

Where AI helps in hiring

StageAI canPeople should
Job advertDraft clear, inclusive advertsCheck accuracy and tone
SourcingSuggest where to advertiseChoose channels
ScreeningSummarise CVs and match evidence to criteriaDecide who progresses
SchedulingBook interviews automaticallyConfirm and prepare
InterviewsSuggest structured questions; take notesInterview and judge
ScorecardsCompile evidence against criteriaMake the final decision
OffersDraft offer lettersApprove and send

Equality law

In the UK, the Equality Act 2010 prohibits discrimination based on protected characteristics such as age, disability, race, sex, religion and others. You're responsible for discriminatory outcomes even if a tool produced them. Similar laws apply in the EU and US.

Data protection

Under UK and EU GDPR you need a lawful basis, transparency about how applications are processed, data minimisation and appropriate retention. Rules restrict solely automated decisions with significant effects; the UK's 2025 reforms relaxed some of these but still require safeguards, including the right to human review. Read using AI tools without breaking GDPR. The UK's ICO has audited AI recruitment tools and published recommendations.

AI-specific rules

  • EU AI Act: AI for recruitment and selection is classed as high-risk, with obligations such as risk management, human oversight and transparency. These have been delayed to 2027–2028, but they show where expectations are heading.
  • US: New York City requires bias audits and candidate notices for automated employment decision tools; other states have rules on AI in interviews and hiring.

How to screen fairly with AI

Write must-haves and nice-to-haves based on the actual job: skills, experience, qualifications that are genuinely required. Avoid proxies that correlate with protected characteristics, such as "graduated in the last 5 years" (age) or "native speaker" (race or national origin) when fluency is what you need.

2. Use evidence-based matching

The safest approach is AI that shows the evidence for each criterion from the CV ("5 years managing payroll at X Ltd"), rather than producing a mysterious score. People can check it, and candidates are judged on what they've done.

3. Keep humans deciding

AI should sort, summarise and highlight. A trained person should decide who progresses, especially for rejections.

4. Hide irrelevant information

Consider removing names, photos, dates of birth and addresses at the screening stage to reduce bias.

5. Test outcomes for bias

Periodically compare pass rates across groups where you lawfully hold monitoring data. Large differences need investigating.

6. Make reasonable adjustments

Offer alternatives for candidates with disabilities, for example if an AI video interview or timed test disadvantages them.

7. Be transparent

Tell candidates in your privacy notice and job adverts how applications are reviewed, including AI use, and how to request human review.

8. Keep records

Document criteria, tool settings, decisions and reasons. You may need them to answer complaints.

Red flags in AI hiring tools

  • Scores with no explanation.
  • Analysis of facial expressions, voice tone or "personality" from video (scientifically weak and high-risk).
  • Training on your past hiring decisions without bias checks.
  • No way to adjust criteria or override results.
  • Vague answers about data storage and security.

Write better job adverts too

Fair hiring starts with the advert: clear requirements, salary range, inclusive language and no unnecessary demands. Read prompt writing for business teams for a job advert template.

A tool built for this

Zelvorm Hire is AI recruiting for small businesses: job adverts, a careers page that appears on Google, CV screening that shows the evidence for each requirement, interview scheduling, scorecards and offers, with people making the decisions.

The short version

AI can save days on hiring by summarising CVs, matching evidence to criteria and handling scheduling. To stay fair and legal, define job-related criteria, use evidence-based matching rather than black-box scores, keep humans making decisions, test for bias, make adjustments, be transparent with candidates and keep records.

Want help designing a fair hiring process? Our consulting team can help, or start with Zelvorm Hire for free.