Imagine a digital judge or clerk that makes decisions but can't explain why or might be accidentally biased against certain people. This work builds a set of 'safety checks' and 'translation tools' that make these AI systems fair and easy for humans to understand. It's like adding a transparent glass wall and a guidebook to a black-box machine so everyone knows the rules are being followed.
Government agencies using AI risk legal penalties and public backlash if their automated decisions are biased or cannot be explained. There is a critical lack of tools to detect and remove this bias while remaining compliant with new laws like the AI Act.
A Bias Detector toolkit, a Virtualized Unbiasing Framework (VUF), and a library of XAI/SAX models for explaining AI decisions. It also includes Blockchain-based governance applications and a catalogue of policy models.
If you are a software provider dealing with strict AI Act compliance for city halls — this project developed a Bias Detector and Virtualized Unbiasing Framework that ensures automated decisions are fair and non-discriminatory.
If you are an auditing firm dealing with the need to prove AI fairness to regulators — this project developed XAI algorithms and a library for explaining AI decisions that turn complex code into human-readable justifications.
If you are a tech firm dealing with trust issues in digital voting or governance — this project developed decentralized Blockchain-based applications that monitor and revise governance rules throughout their lifecycle.
Based on available project data, no commercial pricing or licensing costs are mentioned; the project received a EUR 3,000,000 EU contribution for development.
The project includes 7 industry partners and developed software implementations for monitoring and visualization, suggesting a move toward scalable tools, though full industrial scale is not explicitly confirmed.
Based on available project data, specific licensing terms are not provided, but the project delivers a catalogue of policy models and datasets for reuse in different domains.
The project specifically targets AI Regulation and the AI Act by providing technical measures to mitigate bias and ensure fairness in AI systems.
The project defined interfaces and interactions between toolkits, libraries, and components to ensure different software parts work together.
The consortium is heavily industry-weighted with 7 companies (50% of the 14 partners), including 3 SMEs. This strong commercial presence, combined with partners from 8 different countries, indicates a high focus on practical application and cross-border regulatory alignment rather than purely academic research.
Contact MAGGIOLI SPA in Italy for technical inquiries regarding the Bias Detector toolkit.
Request a technical deep-dive into the XAI library for your compliance roadmap.
