Our Approach
Understanding and addressing bias in AI systems through human-centered measurement and evaluation. Developed by Copenhagen Municipality data scientist, funded by Innovationsfonden Innofounder 2025.
The Challenge
Across Europe, public institutions must now prove that their AI systems treat citizens fairly — yet most organisations have no practical way to measure or document bias hidden deep inside complex models. Without clear, standardised insight, they risk non-compliance, lost trust and costly re-engineering.
Our Solution
FairAI Monitor is a cloud-based dashboard that automatically audits your machine-learning and generative models for input- and output-bias, benchmarks the results against human data, and exports a ready-made EU AI Act compliance report—giving public institutions instant, defensible proof of fairness and a clear path to improvement.
FairAI Baseline Benchmarks
Human vs. AI Fairness
FairAI Monitor
Automated Bias Detection
FairAI Compliance Pack
AI Act Reporting & Risk
FairAI Baseline Benchmarks
Your reference point for what "fair" looks like in practice. Sector surveys and human annotations establish a defensible baseline of human bias.
Sector surveys and human annotations establish a defensible baseline of human bias.
Compare model behavior to real human decisions (e.g., caseworker annotations) to see if AI is better—or worse.
Shared benchmarks enable apples-to-apples comparisons across municipalities and agencies.
Backed by a research partnership with Aarhus and Lund University for method design and baseline building.
FairAI Monitor
A SaaS platform that quantifies bias in your AI. Upload data, analyze input/output bias for classification & regression; generative model support on the roadmap.
Upload data, analyze input/output bias for classification & regression; generative model support on the roadmap.
Uses human-annotated baselines to reveal where models skew across groups (e.g., gender, ethnicity, age).
Dashboard with clear fairness metrics and trends for decision-makers.
FairAI Compliance Pack
Turn technical findings into audit-ready documentation. One-click, AI Act–aligned reports with the right metrics and wording for governance.
One-click, AI Act–aligned reports with the right metrics and wording for governance.
Text-based risk assessment tailored to public-sector requirements.
EU-hosted (Scaleway) deployment for GDPR-compliant evidence and data handling.