Democratic Commons — AI for Democracy

Democratic Commons is an open research and technology initiative exploring how AI can support democracy while respecting democratic principles.

Led by Sciences Po, Sorbonne Université and Make.org, the initiative brings together social science, AI research and real-world civic experimentation to build and share datasets, models, benchmarks and tools for AI that serves democracy.

Why AI for democracy?

AI can make democratic participation more accessible, help citizens navigate complex information and enable collective sensemaking at unprecedented scale. But it can also introduce biases, distort representation and influence democratic processes in ways that are difficult to detect.

Our objective is therefore not simply to use AI for democracy, but to understand how AI should behave when used in democratic contexts — and how we can measure whether it does.

Democratic principles as our foundation

What does it mean for an AI system to work for democracy?

Research led by the Sciences Po team within Democratic Commons provides the conceptual foundation for answering this question. Their work identifies fundamental democratic principles that should guide the design and evaluation of AI systems used in democratic contexts.

These principles form the basis of our work across the consortium: they help us identify potential risks and biases, build evaluation methodologies and benchmarks, and ultimately design AI systems that better serve democratic processes.

• Participation — AI should enable meaningful participation and empower citizens rather than reduce their agency.
• Pluralism — AI should preserve the diversity of opinions and perspectives, including minority and underrepresented voices.
• Deliberation — AI should support informed and constructive exchanges between different perspectives.
• Responsibilities — The use of AI should preserve human responsibility, accountability and the ability to understand its role and influence.
• Agreements & disagreements — AI should help surface both common ground and meaningful disagreement, without artificially creating consensus.

Together, these principles provide a common framework to move from democratic theory to measurable AI behaviour.

Explore our open resources

We translate these principles into concrete research and technical resources, openly shared with the community.

📊 Datasets — Democratic participation data and curated datasets for AI research.

🧪 Benchmarks — Evaluation frameworks and test sets for measuring AI against democratic principles.

🤖 Models — Models developed, adapted or evaluated for democratic use cases.

🛠️ Tools — Open-source components for participation, deliberation and collective sensemaking.

📚 Research — Papers and methodologies connecting democratic principles with AI design and evaluation.

New resources will be added as they are produced by the Democratic Commons consortium and its community.

From research to real-world democracy

Democratic Commons combines expertise in political and social sciences from Sciences Po, AI and computer science research from Sorbonne Université, and large-scale digital participation and civic-tech experimentation from Make.org.

This combination allows us to move between fundamental research, technical development and experimentation in real democratic contexts — and to learn from those experiments to improve both our methods and our technologies.

Join the community

Democratic Commons is designed as an open community, not a closed research project.

We want these resources to be used, challenged and improved by researchers, AI practitioners, civic-tech organizations and democratic institutions.

Use our datasets → Test the benchmarks → Evaluate models → Share results → Contribute new resources

If you are working on AI for democracy, democratic participation, collective intelligence, AI evaluation or related topics, we invite you to explore our resources and contribute to the commons.

About Democratic Commons

Democratic Commons is a French research initiative led by Sciences Po, Sorbonne Université and Make.org, bringing together research, technology and civic experimentation to develop AI that strengthens rather than weakens democracy.

🌐 Democratic Commons website