AI in recruitment: a practical guide for HR teams

In practice, AI supports several distinct steps in hiring:

• Drafting job adverts and job descriptions, based on patterns from past listings

• Answering candidate questions through a company-specific chatbot, before an application is even submitted

• Shortlisting applications by matching CVs against the qualifications a role requires

• Profiling candidates in assessment centres, surfacing hard skills and soft skills through how they respond to tasks

Under the EU AI Act, AI used to screen CVs, score candidates, or shape hiring outcomes is classified as high-risk. High-risk systems carry firm obligations: documented assessment criteria, transparency about how scoring works, and meaningful human oversight at each decision point. Employers must also tell candidates when AI has evaluated their application. The rules apply from December 2027, and non-compliance can cost up to €15 million or 3% of global annual turnover.

GDPR sets a stricter bar today. Since 2018, GDPR Article 22 has given candidates the right not to face a hiring decision made solely by an automated system. In July 2026, EU regulators confirmed that many organisations' automated hiring practices have been breaching that right since 2018 (TechTimes, 2026). The obligation to explain an automated decision to a candidate, or their representative, already exists, and it doesn't shift with the AI Act's later deadline.

For Personio's own AI Talent Screening Agent, this isn't a compliance afterthought. The rule we build to is simple: AI prioritises, the recruiter decides. The Agent scores every application against criteria your team sets, such as experience, location, skills, and education, and shows that reasoning inside the candidate profile. It never rejects a candidate automatically. Hiring decisions shape careers and teams, and that judgement stays with people.

Confidence in AI hiring splits sharply between who's being hired and who's doing the hiring. Greenhouse's 2025 research found that half of candidates think AI has improved hiring overall, mainly through faster screening and scheduling. But a quarter aren't confident in AI hiring systems at all, and 8% say they have no idea what an algorithm is prioritising when it evaluates them. Overall, just 8% of candidates believe AI makes hiring fairer.

That gap sits against rising pressure on recruiting teams: application volumes keep climbing while recruiter capacity doesn't. Matching AI to the right hiring situation matters more than how much AI a team uses overall.

Two things determine where AI helps most: how many people you're hiring, and how much leverage candidates have in the market.

Most teams track time-to-hire, cost-per-hire, and source. All are useful, but none reveal whether the resulting hire performed well, stayed, or strengthened the team.

Without that link, teams keep optimising recruiting against process metrics with no signal on whether the process produces the right people. Connecting hiring, onboarding, performance, and retention data on one platform closes that gap. Teams can then check three to six months after a hire whether someone passed probation, performed strongly through their first year, or struggled to fit in and left.

• Impersonal delivery: candidates can find automated chatbots or generic rejection messages impersonal. Pairing automation with a named human contact protects the candidate experience.

• Limited explainability: many AI systems can't fully explain why they scored a candidate the way they did, making it harder for recruiters to justify exclusions.

• Compliance exposure: under GDPR and the EU AI Act, a system that screens or scores candidates without meaningful human review creates real legal risk, not just a reputational one.

1. Set your hiring goals. Decide whether you're solving for speed, quality, availability, or all three, before you evaluate any tool.

2. Audit your current process. Identify where candidates drop off, where bias creeps in, and where AI could help.

3. Choose the right tool for the job. An applicant tracking system with AI built in, like Personio, covers the whole process; specialist tools cover one step.

4. Prepare your data. Clean, well-structured job descriptions produce better AI outputs, and better decisions.

5. Pilot before you scale. Test the tool on a defined group of candidates or roles, and compare its output against your existing process.

6. Set clear human checkpoints. Decide upfront which decisions AI can support and which always need a person, and tell candidates when AI has been used.

AI in recruitment refers to tools that support hiring tasks such as CV screening, candidate shortlisting, chatbot-led candidate queries, and predicting likely job performance. It speeds up and structures decisions that recruiters ultimately make.

Yes, with conditions. GDPR Article 22 has given candidates the right not to face a fully automated hiring decision since 2018. Using AI to support a recruiter's decision is compliant; letting AI reject a candidate without meaningful human review is not.

From December 2027, AI systems used to screen, score, or rank candidates are classified as high-risk. This brings requirements for documented assessment criteria, transparency about scoring, human oversight at each decision point, and disclosure to candidates when AI has evaluated their application.

No. AI works best on repetitive, high-volume tasks, freeing recruiters to focus on judgement calls and relationship-building. Strategic decisions, such as which channels to prioritise or how to handle a borderline candidate, stay with people.

Less confident than hiring teams often assume. Only 8% of candidates believe AI makes hiring fairer, and a quarter say they aren't confident in AI hiring systems at all (Greenhouse, 2025).