Selected work

Representative Papers

Applied and Filtered: An End-to-End Algorithmic Fairness Audit of A Public Employment Agency

Aug 13, 2026

This study addresses the limitations of prevailing algorithmic fairness assessments, which often focus narrowly on technical metrics while neglecting organizational and societal contexts, thereby failing to uncover systemic biases in real-world deployments. It presents the first end-to-end socio-technical audit of a semi-automated hiring system used by Barcelona’s public employment service from 2017 to 2022, analyzing nearly 500,000 candidate–job pipeline records. Integrating disparate impact ratio (DIR), multi-stage tracking, and intersectional fairness measures across gender, age, and salary levels, the analysis reveals that while overall gender representation appears balanced, women are significantly underrepresented in shortlists for mid-salary positions (DIR = 0.786), non-binary individuals are selected at less than one-third the rate of men, and candidates over 55 are entirely absent. The audit further uncovers process-level biases invisible to model-centric evaluations and highlights critical information asymmetries between vendors and deploying institutions.

0 citationsRead paper

The Eticas AI Risk Taxonomy: Open Infrastructure for Operationalizing AI Audits

Jul 02, 2026

This study addresses the gap between conceptual AI risk frameworks and actionable auditing methodologies by proposing the first end-to-end operationalizable framework. It decouples risk definitions into their manifestation mechanisms and introduces Eticas AI Risk Taxonomy v2.0.0—an open, extensible classification system comprising 76 subcategories—demonstrating a complete pipeline from risk definition through executable testing, quantitative scoring, to risk tiering, exemplified by PII leakage risks. The taxonomy is published under CC BY 4.0 using SKOS/JSON-LD semantic standards, providing stable URIs, calibrated thresholds, and mappings to 18 external frameworks. Empirical evaluation on GPT-4-0314 reveals PII disclosure rates rising to 84% under adversarial prompting, leading to its classification as an E-level systemic risk, thereby validating the framework’s effectiveness and practical utility.

0 citationsRead paper

Continuous Control of Editing Models via Adaptive-Origin Guidance

Feb 03, 2026

Existing diffusion-based editing methods struggle to achieve smooth, continuous control over editing intensity under text guidance. This work proposes an Adaptive Origin Guidance (AdaOr) mechanism that dynamically modulates editing strength during inference by interpolating between identity-conditioned and unconditional predictions. AdaOr effectively addresses the discontinuous transitions inherent in conventional Classifier-Free Guidance when applied to editing tasks. Notably, the method requires neither specialized training datasets nor per-edit optimization, yet enables fine-grained, consistent, and fluid control in both image and video editing. Experimental results demonstrate that AdaOr significantly outperforms existing slider-based intensity control strategies in terms of visual quality and controllability.

0 citationsRead paper

Recent publications

Latest Papers

Applied and Filtered: An End-to-End Algorithmic Fairness Audit of A Public Employment Agency

Aug 13, 2026

This study addresses the limitations of prevailing algorithmic fairness assessments, which often focus narrowly on technical metrics while neglecting organizational and societal contexts, thereby failing to uncover systemic biases in real-world deployments. It presents the first end-to-end socio-technical audit of a semi-automated hiring system used by Barcelona’s public employment service from 2017 to 2022, analyzing nearly 500,000 candidate–job pipeline records. Integrating disparate impact ratio (DIR), multi-stage tracking, and intersectional fairness measures across gender, age, and salary levels, the analysis reveals that while overall gender representation appears balanced, women are significantly underrepresented in shortlists for mid-salary positions (DIR = 0.786), non-binary individuals are selected at less than one-third the rate of men, and candidates over 55 are entirely absent. The audit further uncovers process-level biases invisible to model-centric evaluations and highlights critical information asymmetries between vendors and deploying institutions.

0 citationsRead paper

The Eticas AI Risk Taxonomy: Open Infrastructure for Operationalizing AI Audits

Jul 02, 2026

This study addresses the gap between conceptual AI risk frameworks and actionable auditing methodologies by proposing the first end-to-end operationalizable framework. It decouples risk definitions into their manifestation mechanisms and introduces Eticas AI Risk Taxonomy v2.0.0—an open, extensible classification system comprising 76 subcategories—demonstrating a complete pipeline from risk definition through executable testing, quantitative scoring, to risk tiering, exemplified by PII leakage risks. The taxonomy is published under CC BY 4.0 using SKOS/JSON-LD semantic standards, providing stable URIs, calibrated thresholds, and mappings to 18 external frameworks. Empirical evaluation on GPT-4-0314 reveals PII disclosure rates rising to 84% under adversarial prompting, leading to its classification as an E-level systemic risk, thereby validating the framework’s effectiveness and practical utility.

0 citationsRead paper

Continuous Control of Editing Models via Adaptive-Origin Guidance

Feb 03, 2026

Existing diffusion-based editing methods struggle to achieve smooth, continuous control over editing intensity under text guidance. This work proposes an Adaptive Origin Guidance (AdaOr) mechanism that dynamically modulates editing strength during inference by interpolating between identity-conditioned and unconditional predictions. AdaOr effectively addresses the discontinuous transitions inherent in conventional Classifier-Free Guidance when applied to editing tasks. Notably, the method requires neither specialized training datasets nor per-edit optimization, yet enables fine-grained, consistent, and fluid control in both image and video editing. Experimental results demonstrate that AdaOr significantly outperforms existing slider-based intensity control strategies in terms of visual quality and controllability.

0 citationsRead paper