By Johannes Sundlo — AI & Future of Work Advisor. I help leaders and HR teams turn AI into adoption that sticks, through keynotes, workshops and change programs.
Recruiting is one of the fastest places to get value from AI — and one of the easiest to get wrong. Used well, it removes hours of repetitive work and sharpens your decisions. Used carelessly, it bakes in bias and erodes candidate trust. This guide is the practical line between the two.
Where AI genuinely helps in recruiting
- Drafting: first-draft job ads in your voice, outreach messages, structured interview guides tied to the role’s criteria.
- Screening support: summarising long free-text applications and comparing candidates against your stated criteria — as input, never the final call.
- Faster admin: scheduling copy, interview-note summaries, turning a debrief into a structured scorecard.
- Sourcing research: summarising a market or role, drafting boolean searches, prepping talking points.
Where AI backfires — and how to avoid it
- Automated rejection. Never let a model auto-reject candidates. Use it to surface and structure, not to decide.
- Hidden bias. AI can amplify patterns in historical data. Keep humans accountable for every decision and check outcomes for disparity.
- Generic, soulless output. A job ad that reads like every other AI job ad repels good people. Always edit for your voice and specifics.
- Candidate-data risk. Be deliberate about what personal data goes into which tool, and tell candidates how AI is used.
A practical way to introduce AI into hiring
- Pick one high-frequency task (job ads or interview guides) and build a reusable prompt with your tone and criteria.
- Keep a recruiter in the loop on every output — AI drafts, a person decides.
- Write down a one-page rule for candidate data and transparency before you scale.
- Measure time saved and quality (do shortlists get better?), then expand to the next task.
The legal landscape: EU AI Act, NYC Local Law 144 and what they mean for you
Recruiting is one of the most regulated corners of AI, and the rules are no longer theoretical. Two matter most in 2026:
- EU AI Act. AI used for recruitment and selection, screening applications, evaluating candidates, is classed as high-risk under Annex III. From August 2026 that brings concrete obligations: documented risk management, quality controls on training data, human oversight of decisions, and transparency towards candidates. It applies to anyone hiring in the EU, regardless of where the tool vendor sits.
- NYC Local Law 144. If you use an automated employment decision tool for roles in New York City, you need an annual independent bias audit, you must publish a summary of the results, and you must notify candidates that the tool is used. It has been enforced since 2023 and has become the template other US states copy.
- More on the way. Illinois regulates AI video interviews, Colorado’s AI Act takes effect in 2026, and several other jurisdictions have bills in progress. The direction is the same everywhere: transparency, human accountability, and proof that you checked for bias.
The practical response is not to avoid AI, it is to run a tighter process: keep an inventory of every AI touchpoint in your funnel, ask vendors for their bias-audit or conformity documentation before you sign, keep a human making the final call, and tell candidates how AI is used. Teams that do this are compliant almost by default, whatever jurisdiction comes next.
A worked example: job ads in a third of the time
The pattern I see most often in workshops with talent teams: a recruiter spends 60 to 90 minutes on a job ad, staring at a blank page, borrowing from an old ad, then fixing the borrowed bias baked into it. With a reusable prompt that carries your tone of voice, the role’s actual success criteria and a “what we will not write” list, the first draft takes minutes and the recruiter spends 20 minutes editing instead of an hour writing. For a team publishing ten ads a month, that is roughly a working day back, every month, from one prompt. And the ads get better, because the editing time goes into specifics that attract the right applicants instead of into producing text.
The same logic applies down the funnel: structured interview guides from the criteria, debrief notes into scorecards, outreach that sounds like you. Which tools to do it with? Start with the ones you already have (Copilot, ChatGPT or Claude with your templates), then look at dedicated recruiting tools in the AI tools for HR list once the workflow is proven.
Common questions about AI in recruiting
Can AI screen candidates?
It can help summarise and structure applications against your criteria, but it should never make the final accept/reject decision. Keep a human accountable to avoid bias and legal risk.
Is it legal to use AI in hiring?
Generally yes, but rules are tightening (transparency, anti-discrimination, and in some regions specific AI-in-hiring laws). The safe posture: humans decide, document your process, check for disparate impact, and be transparent with candidates.
Does the EU AI Act apply to recruiting?
Yes. AI systems used for recruitment, screening and candidate evaluation are classed as high-risk under Annex III, with obligations on risk management, human oversight and candidate transparency applying from August 2026. If you hire in the EU, it applies to you regardless of where the tool vendor is based.
Recruiting is one slice of a bigger shift. For the full picture, see AI for HR and the hub guide on AI adoption that sticks.
Bring practical AI to your talent team
A workshop where your recruiters leave with working prompts and a clear, safe way to use AI in hiring. Let’s set it up.
AI in recruiting: where it helps vs. where it backfires
| Task | Where AI helps | Where it backfires |
|---|---|---|
| Job ads | First drafts in your voice | Generic, soulless copy |
| Screening | Summarise & structure as input | Auto-rejecting candidates |
| Decisions | Surface evidence for a human | Making the final call (bias & legal risk) |
| Candidate data | — | Pasting sensitive data into public tools |
The data behind this: see the State of AI in HR report — original survey data on the AI adoption gap.
