AI Governance & Transparency

Clapwork is committed to the responsible use of AI in hiring. We believe that AI should augment human decision-making, not replace it. Our approach is guided by the principles of transparency, fairness, accountability, and human oversight.

This document describes how AI is used across the Clapwork platform, the safeguards we have in place, and the rights available to candidates and employers.

Clapwork uses AI in the following areas:

• AI Interview Question Generation: When a hiring manager creates an interview from a skills list, AI generates relevant, structured questions calibrated to the required proficiency level.
• Response Evaluation & Scoring: AI evaluates candidate responses to produce skill proficiency ratings, holistic match scores, and soft-skill assessments. Scoring uses documented rubrics aligned to the skills being evaluated.
• Language Proficiency Assessment: AI analyzes spoken and written responses to assign CEFR-aligned language proficiency levels (A1-C2).
• Resume Scoring: AI compares resume content against job description requirements to generate an alignment score.
• Candidate Predictions: AI generates optional predictive signals including retention likelihood, attrition risk, and placement confidence. These are advisory only and never used as sole decision factors.
• Proctoring & Integrity: AI monitors assessment sessions for integrity signals such as tab-switching, external tool use, and anomalous patterns.
• Recommendations: AI may suggest interview questions, propose candidate rankings, or recommend next steps in the hiring pipeline.

We take active steps to identify and mitigate bias in our AI systems:

• Structured Assessment Design: All candidates for the same role answer the same questions under the same conditions, reducing unstructured interview bias.
• Skills-Based Scoring: Evaluation focuses on demonstrated competency against documented criteria rather than proxies like education pedigree or employer prestige.
• Anonymized Review: Our bias blocker feature allows hiring managers to review candidate scores without seeing names, photos, or demographic information.
• Regular Auditing: We conduct regular audits of scoring distributions across demographic groups to identify and address disparate impact.
• Human Override: All AI-generated scores can be reviewed and overridden by human decision-makers.

AI-generated assessments on Clapwork are always subject to human oversight:

• AI scores are presented as decision-support tools, not automated hiring decisions.
• Hiring managers retain full authority over all hiring decisions.
• Candidates may request human review of AI-generated assessments.
• AI-generated predictions are labeled as advisory and never used as automated gatekeepers.

Candidates assessed through Clapwork's AI systems have the following rights:

• Right to Information: Candidates are informed before an assessment that AI will be used in the evaluation process.
• Right to Explanation: Candidates can request an explanation of how their assessment was scored, including the criteria used and the weight given to different factors.
• Right to Human Review: Candidates can request that a human reviewer evaluate their assessment independently of AI scores.
• Right to Object: Candidates can object to solely automated decision-making that significantly affects them, as provided under applicable data protection laws.
• Right to Data Access: Candidates can request access to their assessment data, recordings, and scores through the platform or by contacting support.

Clapwork may use aggregated, de-identified data to improve AI model accuracy and reduce bias. Individual assessment responses are never shared with third parties for model training purposes without explicit consent.

Users can opt out of AI model improvement through their AI Controls settings page.

Clapwork classifies its AI-powered recruitment tools as high-risk AI systems under the EU AI Act (Article 6, Annex III, point 4). As such, we maintain:

• A risk management system proportionate to the level of risk.
• Data governance practices ensuring training data quality and representativeness.
• Technical documentation describing system capabilities and limitations.
• Record-keeping of system logs for traceability.
• Transparency obligations to users and affected persons.
• Human oversight mechanisms enabling effective oversight by natural persons.
• Accuracy, robustness, and cybersecurity measures appropriate to the system's purpose.

For questions about our AI governance practices, to exercise your rights as a candidate, or to report concerns about AI-generated assessments, contact our AI Ethics team.