The short answer
Under the EU AI Act, AI systems used to recruit or select people, for example to filter applications or evaluate candidates, are high-risk (Annex III, point 4). After the 2026 Digital Omnibus, the high-risk obligations for these systems apply from 2 December 2027. Employers that use them must follow the provider's instructions, assign trained people to oversee them, keep logs and tell candidates. Emotion recognition in the workplace has been banned since February 2025.
Key takeaways
- AI used to place targeted job ads, filter applications or evaluate candidates is high-risk under Annex III, point 4 of the AI Act.
- The Digital Omnibus on AI (Regulation (EU) 2026/1744) moved the start date for these obligations from 2 August 2026 to 2 December 2027.
- Some rules already apply: since 2 February 2025, AI that infers emotions in the workplace is prohibited, outside narrow medical or safety uses.
- Most employers are "deployers", not "providers". Their duties are human oversight by competent people, monitoring, log retention and informing workers and candidates.
- The Act reaches beyond the EU: it can apply when an AI system's output is used in the EU, for example to assess candidates for an EU-based role.
- The US has its own patchwork, including New York City's bias-audit law, Illinois's notice rule and Colorado's rewritten AI law.
On this page9 sections
What the AI Act says about hiring
The AI Act treats AI in recruitment as high-risk because it affects people's access to work. Annex III, point 4 lists, under "employment, workers' management and access to self-employment", AI systems intended to be used for the recruitment or selection of people, in particular to place targeted job advertisements, to analyse and filter job applications, and to evaluate candidates. It also covers AI used to make decisions about promotion, termination, task allocation and performance monitoring [1].
In practice, that captures many tools hiring teams already use or are evaluating:
- CV or application screening that ranks, scores or filters candidates
- Automated assessment of video interviews, tests or written answers
- Matching tools that decide which candidates a recruiter sees
- Targeted job-ad delivery that decides who sees a vacancy
There is a narrow exception. Under Article 6(3), an Annex III system is not high-risk if it doesn't pose a significant risk, for example because it only performs a narrow procedural task or improves the result of a human activity already completed. That exception never applies when the system profiles people [1]. Ranking candidates on their personal characteristics is profiling, so most screening tools won't qualify.
The timeline after the Digital Omnibus
The high-risk rules for hiring tools now apply from 2 December 2027. The AI Act entered into force on 1 August 2024 with staggered dates. In July 2026 the EU adopted the Digital Omnibus on AI, Regulation (EU) 2026/1744, which postponed the high-risk obligations for Annex III systems from 2 August 2026 to 2 December 2027 [2]. Many articles written in 2025 still give the old date.
| Date | What applies | Relevance to hiring |
|---|---|---|
| 1 August 2024 | The AI Act enters into force | Start of the transition |
| 2 February 2025 | Prohibited practices (Article 5) and AI literacy (Article 4) | Workplace emotion recognition banned; staff using AI need appropriate AI literacy |
| 2 August 2025 | Rules for general-purpose AI models, governance and penalties | Mainly affects model providers |
| 2 August 2026 | Transparency obligations (Article 50) and most remaining provisions | Disclosure duties for some AI interactions and generated content |
| 24 July 2026 | Digital Omnibus on AI published in the Official Journal | Moves the high-risk dates |
| 2 December 2027 | High-risk obligations for Annex III systems | Recruitment, selection and worker-management AI |
| 2 August 2028 | High-risk obligations for AI in products under Annex I | Mostly regulated products, not HR software |
The Omnibus also reworded the AI literacy duty in Article 4 into an obligation to take measures for staff AI literacy [2]. The direction is unchanged: people who use AI at work should understand what it does and where it fails.
What's already banned: emotion recognition at work
Since 2 February 2025, the AI Act has prohibited AI systems that infer the emotions of people in the workplace or in education, except for medical or safety reasons (Article 5(1)(f)) [1]. For hiring, that rules out tools that claim to read a candidate's enthusiasm, honesty or stress from their face or voice in a video interview. Prohibited practices carry the Act's highest fines: up to €35 million or 7% of worldwide annual turnover, whichever is higher (Article 99) [1].
Provider or deployer: which one are you?
Most employers are deployers. The AI Act splits duties between the provider (who develops the AI system and puts it on the market) and the deployer (who uses it under its own authority). If you buy a screening tool from a vendor, the vendor is the provider and you are the deployer. If you build your own screening model, you may be both.
Under Article 26, deployers of high-risk systems must [1]:
- Use the system according to the provider's instructions for use.
- Assign human oversight to people with the necessary competence, training and authority, and give them support.
- Make sure input data they control is relevant and sufficiently representative for the intended purpose.
- Monitor the system and tell the provider, and in some cases the authorities, about serious incidents or risks.
- Keep the logs the system generates for at least six months, unless other law says otherwise.
- Inform workers' representatives and affected workers before putting a high-risk system into use at the workplace.
- Tell people when a high-risk system is used to make, or help make, decisions about them. For hiring, that means candidates.
Article 14 sets the design side of human oversight: high-risk systems must let the people overseeing them understand the system's limits, avoid over-relying on its output, and decide not to use it, override it or stop it [1]. Article 86 gives people affected by certain decisions based on high-risk systems a right to a clear explanation of the AI's role [1].
Providers carry the heavier load: risk management, data governance, technical documentation, accuracy and robustness, conformity assessment and registration. If you buy hiring AI, ask your vendor how they plan to meet those obligations before December 2027.
Is a résumé keyword filter "AI"?
Not always. The Act defines an AI system as a machine-based system that infers from its input how to generate outputs such as predictions, recommendations or decisions. Recital 12 says the definition should not cover systems based on rules defined solely by people to execute operations automatically [1]. The European Commission published guidelines on the definition in February 2025 [3]. A filter that only applies a rule you wrote, such as "must have a work permit for Germany", is likely outside it. A tool that learns to score or rank candidates is likely inside it. Borderline cases need a legal view.
Does the AI Act apply to companies outside the EU?
It can. The Act applies to providers that place AI systems on the EU market and to deployers located in the EU. It also applies to providers and deployers outside the EU where the output of the AI system is used in the EU (Article 2) [1]. If a US company uses an AI tool to screen candidates for a role based in Poland or Germany, that output is used in the EU, so plan as if the Act applies.
Data protection applies too. Under the GDPR, people have the right not to be subject to a decision based solely on automated processing that significantly affects them, with limited exceptions and safeguards (Article 22) [7]. A fully automated rejection of a candidate in the EU raises GDPR questions today, regardless of the AI Act dates.
Beyond the EU: AI hiring rules in the US
The US has no single federal AI hiring law, but state and city rules are growing:
| Jurisdiction | Rule | What it requires (summary) | Status |
|---|---|---|---|
| New York City | Local Law 144 | A bias audit of automated employment decision tools before use, a public summary of results and notice to candidates | Enforced since July 2023 [4] |
| Illinois | Public Act 103-0804 (HB 3773), amending the Human Rights Act | Using AI in a way that discriminates, or without notice, in employment decisions is a civil-rights violation | In effect since 1 January 2026 [5] |
| Colorado | SB 26-189 (repeals and re-enacts the Colorado AI Act) | Notice and disclosure, explanations after adverse decisions, rights to access data and request human review | Signed May 2026; key obligations from 1 January 2027 [6] |
The common thread across the EU and the US is the same: tell people when AI is involved, check it for bias, and keep a person able to review and overturn its output.
A practical checklist for employers
Use this to get ready before December 2027:
- Inventory every tool in your hiring process that ranks, scores, filters or summarises candidates, including features inside your applicant tracking system.
- Classify each one. Is it AI under the Act's definition? Is it used for recruitment or selection? Does it profile people?
- Remove anything that infers emotions from faces, voices or text in interviews. That's already prohibited in the EU.
- Ask vendors for their AI Act plan: instructions for use, documentation, conformity assessment and log access.
- Name the people who oversee each tool, train them, and give them authority to override it.
- Write the candidate notice: what the tool does, at which step, and how to ask for human review.
- Keep logs for at least six months and document decisions, especially rejections.
- Check US exposure: NYC bias audits, Illinois notice and Colorado's 2027 obligations.
For a wider view of where AI fits in the process, see human-in-the-loop hiring.
How we use AI at Tailored Minds
We use AI the way the regulation points: it assists, people decide. On Tailored Minds, AI helps our team search wider. It reads skills rather than keywords, and AI-generated talent reports are advisory. A person reviews and curates every shortlist before a company sees it, and the Highly Qualified badge comes from human review, never from an algorithm alone. Companies see only the candidates we've deliberately shared for their role, and the hiring decision stays with them. You can read more about how we keep quality high and how talent profiles are reviewed. For the full hiring process for AI roles, see our guide to hiring AI engineers.
This describes our practice. It is not a claim of certification or legal compliance on anyone's behalf.
Frequently asked questions
Is AI CV screening high-risk under the EU AI Act?
When do the AI Act's hiring rules apply?
Does the EU AI Act apply to US companies hiring in Europe?
What does human oversight mean in practice?
Do candidates have to be told that AI is used?
Sources
- Regulation (EU) 2024/1689 (Artificial Intelligence Act): Articles 2, 5, 6, 14, 26, 86, 99, Annex III and Recital 12, EUR-Lex
- Regulation (EU) 2026/1744 (Digital Omnibus on AI), EUR-Lex
- The Commission publishes guidelines on AI system definition, European Commission
- Automated Employment Decision Tools (Local Law 144), NYC Department of Consumer and Worker Protection
- Public Act 103-0804 (HB 3773), Illinois General Assembly
- SB26-189: Automated Decision-Making Technology, Colorado General Assembly
- Regulation (EU) 2016/679 (General Data Protection Regulation), Article 22, EUR-Lex
About the author

Bruno Caram
Co-Founder & Board Member at Tailored Minds
Bruno Caram is a co-founder and board member of Tailored Minds. He builds relationships with venture capital firms and the companies they back, and works with founders on how to build their engineering teams.
Share this article
