The Horizon Europe-funded project MAMMOth (Multi-Attribute, Multimodal Bias Mitigation in AI Systems) will officially conclude on the 30th of October 2025, after three years of pioneering work to make artificial intelligence (AI) fairer, more inclusive, and more accountable.
MAMMOth, a consortium of leading universities, research centres, SMEs, and civil society organisations across Europe, was committed to reducing gender, racial, and other forms of bias in AI systems. Its outcomes provide policymakers, technology developers, and civil society actors with concrete tools, knowledge, and strategies to embed fairness at the centre of AI innovation.
Principal Outcomes
- MAI-BIAS Toolkit – An open-source platform enabling developers and researchers to detect, understand, and mitigate bias in datasets and AI models. It offers dashboards, bias detection metrics, model exploration, and transparency functions, now available via the AI-on-Demand platform.
- FairBench & Libraries – Newly developed libraries for fairness-aware machine learning, applied to bias detection in finance, face recognition, and academic citations.
- Scientific Publications – High-impact contributions in fields including computer vision, AI ethics, fairness in machine learning, and law, disseminated through leading journals and international conferences.
- Training & Public Engagement – More than 30 workshops, five webinars, and the MAMMOth Talks podcast series engaged thousands of citizens, researchers, and practitioners. Public exhibitions such as Digital Dilemmas in Thessaloniki and Groningen raised awareness of AI’s hidden biases.
- Policy Recommendations – Three briefs addressed to EU policymakers, civil society, and AI system developers set out practical proposals to align technical solutions with the EU AI Act and to reinforce fairness standards.
- Extensive Outreach – Over 6,600 website visitors, 1,151 social media followers, and more than 12,000 organisations reached through targeted campaigns, significantly surpassing initial outreach targets.
Significance and Legacy
MAMMOth has transformed approaches to AI fairness by moving beyond single-attribute methods to address intersectional and multimodal bias. The project advanced state-of-the-art research while also delivering resources with tangible applications in finance, identity verification, education, and other sectors.
Inclusive co-creation workshops ensured that underrepresented voices informed the development of tools and methodologies, embedding fairness in AI in a context-sensitive and socially responsible manner.
The project’s legacy will continue through its open-source tools, academic resources, and policy frameworks, which remain freely accessible via the AI-on-Demand platform, GitHub, and the official website mammoth-ai.eu. These resources will further support developers, regulators, and communities in building trustworthy, fairness-aware AI systems.
MAMMOth is a Horizon Europe Research and Innovation Action funded by the European Union under Grant Agreement ID 101070285. The project was coordinated by the Centrer for Social Innovation (CSI, Cyprus) with partners across Europe.
For further information, visit: www.mammoth-ai.eu
