ABOUT THE INITIATIVE
An independent response to global concerns about AI
AI BIAIS INITIATIVE was born from a simple conviction: facing warnings from scientists, international institutions, and citizens, a structured, rigorous, and open response was needed.
GENESIS
Why this initiative
For several years, leaders of major AI laboratories, researchers, the Pope, NGOs, and international institutions like UNESCO, the OECD, and the UN have been warning about the dangers of current developments in artificial intelligence.
These warnings are legitimate and converging. But they often remain at the stage of observation, without offering concrete, citable, and open tools for action.
AI BIAIS INITIATIVE aims to fill this void by proposing a common reference framework for AI biases: 250 documented biases, 68 global Regulations, 21 analytical views. All accessible via an open, citable, and sustainable REST API.
A global reach
AI biases have no borders. This initiative is designed to be used everywhere, by everyone, in all languages.
Scientific rigor
Each bias is documented with its Sources, detection metrics, alert thresholds, and citable via DOI.
Total openness
Data published under CC-BY 4.0 license. No barriers, no access conditions. Knowledge must circulate.
FOUNDER
Led by an expert in the field

Mohamadou SIDIBE
Chief Data Officer · Expert en Governance des Données & IA
As a Chief Data Officer and expert in Data and AI Governance, I have seen on a daily basis how algorithmic biases constitute a major societal issue, and how much tools were lacking to document, classify, and process them rigorously.
AI BIAIS INITIATIVE is the answer I wanted to provide: an open, structured, citable, and useful knowledge base, so that Researchers, policymakers, regulators, and developers can finally speak the same language about AI biases.
I also brought the Framework AEXIA, an integrated governance framework for explainable AI that articulates explainability, organizational maturity, and deployed system maturity.
VISION
A fair, explainable, responsible AI
AI is not neutral. It reflects the choices, data, and social structures that produced it. Recognizing its biases is not an admission of weakness - it is the first step towards a truly trustworthy AI.
Equity
No AI system should produce systematic discrimination. Biases must be identified, measured, and corrected.
Explainability
A system that makes decisions about humans must be able to explain them. This is a democratic requirement, not just a technical one.
Responsibility
Behind every algorithm, there are human choices. These choices must be traced, audited, and owned.
VALUES
Our commitments
🌐 Openness
All data is published under CC-BY 4.0 license. Everyone can view, cite, and reuse it.
🔬 Rigor
Each bias is backed by scientific Sources, detection metrics, and documented alert thresholds.
🌍 Universality
AI biases have no borders. The database covers 68 global Regulations and 5 continents.
🤝 Cooperation
The initiative is open to contributions: researchers, NGOs, institutions, citizens. Together, we go further.
Join the movement
Researchers, policymakers, NGOs, institutions, citizens - this initiative is yours. Together, let's build a trustworthy AI.