A recruiter pasting a job description into an AI assistant is not facing the same regulatory exposure as a business using software to rank every applicant before a human sees their CV. Yet both activities may sit within the same hiring process. That distinction is shaping EU AI Act hiring trends across European SMEs: less interest in AI as a stand-alone recruiting tool and more attention on where it influences a decision about a person.
For lean HR teams, the challenge is not to become AI lawyers. It is to keep recruitment efficient while being able to explain which tools are in use, what they do, who checks their output and how candidates are treated fairly. The organisations getting ahead are building those habits now, rather than waiting for a vendor questionnaire or candidate complaint to expose gaps.
Why the EU AI Act changes hiring decisions
The EU AI Act takes a risk-based approach. In recruitment, systems intended to place targeted job advertisements, analyse and filter job applications or evaluate candidates can fall into the high-risk category. The rules for high-risk systems are scheduled to apply from 02 August 2026, although organisations should follow implementation guidance and assess their own use cases carefully.
This does not mean every use of AI in HR is high risk. Using a tool to improve the wording of a job advert, summarise interview notes or generate a first draft of a rejection email may be lower risk when a person remains responsible for the final content and decision. The purpose of the system matters, as does the way your team actually uses it.
There is a clear line that HR leaders should not cross. The Act prohibits certain AI practices, including emotion recognition in the workplace, subject to limited exceptions. A tool that claims to infer confidence, honesty or engagement from a candidate’s voice, face or behaviour should trigger serious concern. The commercial promise may be attractive. The legal, ethical and reputational cost is usually not.
EU AI Act hiring trends to watch
Hiring teams are moving from automation to assisted judgement
For years, recruitment technology has promised speed through automatic scoring and filtering. The direction of travel is now more measured. SMEs are more likely to use AI to reduce administrative work – drafting role profiles, creating interview question banks, summarising structured feedback and answering routine candidate queries – while keeping assessment and selection with trained people.
This is partly a compliance response, but it is also a quality decision. Hiring is contextual. A model may identify matching terms in a CV, but it cannot reliably understand a candidate’s non-linear career, transferable experience or circumstances that deserve reasonable adjustment. Human oversight should be meaningful, not a nominal click at the end of an automated workflow.
The trade-off is speed. A fully automated shortlist can look efficient when vacancies are high volume. But if no one can explain why candidates were screened out, that efficiency can quickly turn into rework, disputes and damaged employer trust.
Procurement questions are becoming more specific
HR teams increasingly need better answers from recruitment software suppliers. “Our platform uses responsible AI” is not enough. Before adopting or renewing a tool, ask what data it processes, whether it makes recommendations or decisions, how those outputs are generated and what controls the customer can configure.
For a high-risk use case, the supplier’s obligations are substantial, but the deploying organisation still has responsibilities. Employers need to use the system according to instructions, provide appropriate human oversight, monitor its operation and keep relevant logs where required. They also need to meet wider GDPR duties. The AI Act does not replace data protection law, equality law or employment law.
A practical supplier review should also cover data residency, sub-processors, retention periods, model training arrangements and whether candidate data is used beyond your organisation’s service. For European SMEs, these questions are easier to manage when core HR data is not scattered across separate recruiting, onboarding and performance tools.
AI literacy is becoming an HR operating requirement
Since 02 February 2025, organisations that provide or use AI systems have needed to take measures to ensure sufficient AI literacy among staff dealing with those systems. There is no single mandatory training course that fits every business. What is sufficient depends on the tool, the risks and the people using it.
For a recruitment team, useful literacy is practical. Recruiters and hiring managers should know when an AI output is a suggestion rather than evidence, how bias can appear, when to escalate an unexpected result and why they must not paste sensitive candidate information into an unapproved public tool. They should also understand the boundaries of their authority when an automated recommendation conflicts with their professional judgement.
A short, role-specific policy and regular training will usually be more valuable than a generic annual presentation. Record who has been trained, what tools are approved and where employees can seek advice. This makes good operational sense even where a particular hiring tool is outside the high-risk category.
Candidates expect clearer explanations
Transparency is becoming a recruitment differentiator. Candidates already expect to know how their data will be used. As AI becomes more visible, they will also ask whether software influenced shortlisting, interview scheduling or assessment.
Clear communication does not require revealing confidential scoring logic. It does mean giving candidates understandable information about the role AI plays, how they can request support or reasonable adjustments and how they can raise a concern. This is especially relevant where an organisation operates across several countries and must keep its candidate experience consistent.
The best approach is not to add a dense paragraph of legal text to every application form. Build a concise explanation into the candidate privacy information, then make sure the process described matches reality. If managers override rankings, retain interview notes or use structured scorecards, say so accurately.
A practical control framework for smaller HR teams
You do not need a separate governance department to improve recruitment AI controls. Start with a simple register of every AI-enabled tool used in the hiring journey, including tools bought directly by HR and general-purpose assistants used informally by managers. Note the purpose, users, data involved, supplier, countries affected and whether the output can influence a candidate’s progress.
Then separate low-impact support from decision-influencing use. A writing assistant for job adverts needs clear data handling rules and human review. A CV screening or assessment system needs closer scrutiny: supplier documentation, testing, oversight procedures, data protection assessment and a clear route for staff to challenge poor outputs.
Next, map the actual workflow rather than the intended one. Ask a recruiter to show how a vacancy is created, candidates are screened, interviews are assessed and decisions are recorded. Informal workarounds often create the biggest risk. For example, a manager may export candidate profiles to an unapproved AI service because the approved recruitment system is slow or fragmented.
Finally, create ownership. HR should own the process and candidate experience, but it should not carry every decision alone. Include the relevant business manager, IT or security lead, data protection contact and legal adviser where the use case warrants it. Review the register when a supplier introduces new AI functionality, not only at contract renewal.
What this means for your HR technology stack
The most sustainable hiring trend is consolidation with control. When recruitment, onboarding, employee records and learning sit in disconnected systems, it becomes difficult to know where personal data has travelled, which AI tools can access it and whether retention rules are being followed.
A enkel HR-platform will not remove the need for judgement or governance. It can, however, make approved workflows, access controls and audit trails easier to manage for a small HR team. C2 is designed around that practical reality: one European HR environment for the processes that otherwise become fragmented as a business grows.
Be wary of treating compliance as a reason to stop using AI altogether. Used carefully, it can give recruiters more time for candidate conversations, stronger interview preparation and consistent administration. The aim is not fewer tools at any cost. It is a hiring process your team can explain, supervise and improve with confidence.
