AI transparency
Last updated: October 8, 2026
Draft document: information in square brackets must be completed before going live.
A high-risk AI system, and we treat it as one
The EU Artificial Intelligence Act (EU 2024/1689) classifies systems used to recruit and assess candidates as high-risk (Annex III). CandidAct is designed to meet these requirements.
Informing candidates
Before they start, every candidate is told that AI analyses their application, their simulation and their voice interview, and how this information is used.
Human oversight
AI produces assessments and summaries. The hiring decision is always made by a human, who has access to everything: the candidate's submission, the rubric, the scores and the transcript.
No fully automated rejection
No application is screened out by AI without human involvement. Every candidate gets the chance to complete the simulation.
Right to an explanation
Every candidate receives feedback on their simulation and can ask how they were assessed by writing to contact@candidact.com.
Fairness and bias monitoring
All candidates for the same role take the same exercise and are scored against the same rubric. The criteria are explicit and approved by the recruiter, and assessments are reviewed regularly to detect potential bias.