The AI Act, the European regulation on artificial intelligence, now frames the use of AI in recruitment. Good news for HR teams: it is not a ban, it is a framework. And its timeline has just shifted.
The timeline has changed
- 1 August 2024: the regulation entered into force.
- 2 February 2025: the prohibited practices already apply.
- 2 August 2025: the rules for general-purpose AI models.
- Recruitment is classified as "high-risk" (Annex III). Its obligations were set for August 2026, but the "Digital Omnibus" agreement pushed them back to 2 December 2027.
The trap to avoid: this delay only covers the "high-risk" obligations. The bans are already in force.
What is already banned (since February 2025)
- Emotion recognition in the workplace and in interviews is banned. A tool cannot infer a candidate's emotional state from their face, voice or biometric signals (outside medical or safety cases). Penalties reach up to 35 million euros or 7% of global turnover.
- Social scoring of individuals is banned.
On top of this, the GDPR already applies: its Article 22 gives candidates the right not to be subject to a fully automated decision. Human involvement therefore remains mandatory on the final decision.
What "high-risk" will require (from December 2027)
For AI systems used in recruitment, the regulation will require, among other things:
- effective human oversight at every decision step,
- transparency toward candidates (knowing that an AI is involved),
- data governance and bias control,
- logging of processing (logs kept for at least six months),
- technical documentation, and for some deployers, a fundamental rights impact assessment.
Why prepare now
- The bans already apply: a tool that "reads emotions" in an interview is unlawful today.
- Compliance is not built overnight: human oversight, traceability and bias control take time.
- Candidates pay more and more attention to how they are assessed. Transparency becomes an advantage, not a constraint.
Evalyo's approach: compliant by design
At Evalyo, the principle is simple: AI structures, you decide. AI offloads the repetitive tasks (test generation, CV scoring, reports), but the human keeps control of every decision.
In practice:
- No emotion recognition: we assess answers and skills, not faces.
- Human oversight: evaluations are decision aids, never automatic decisions.
- Traceability: every test and interview produces a structured, auditable report.
- Bias control: scoring is based on criteria you define, not on name, school or photo.
The AI Act does not penalize AI in recruitment. It penalizes AI without safeguards. That is exactly the philosophy Evalyo was built on.
👉 Discover Evalyo and structure your hiring while keeping people at the center.