Democracy x AI cohort, fellowships, and an experimental incubator

Democracy x AI cohort, fellowships, and an experimental incubator is sponsored by Mozilla Foundation. The Mozilla Foundation channels funding through its Democracy x AI cohort, fellowships, and an experimental incubator.

3. /Mozilla Foundation Incubator 4. /Democracy x AI Cohort 3.

Mozilla Foundation Incubator 4. /Democracy x AI Cohort A better, more democratic tech future isn’t just possible. It’s still ours to build.

## Imagine AI that’s built so more people have a say in what’s next. We’ve been sold the idea that a better future is more efficient, moves faster, and is one where human judgment can be engineered away. The logic is all around us.

It’s in the technology we use, the devices we wear, and the systems that shape our lives. But**Democracy isn’t meant to be optimized. ** The messiness and friction are the point.

Disagreement and compromise aren’t obstacles that need fixing. They’re essential to how democracy works. The Democracy x AI cohort is founded on the hypothesis that, if built differently, AI could strengthen democracy rather than erode it.

Global by design, this cohort reflects the understanding that technology doesn’t emerge from a vacuum: its politics shape its possibilities. We curate the cohort so ideas move across borders, as they should. These ten projects are turning imagination into infrastructure, creating new tools and systems that keep human judgment at the center of democratic life.

## Explore What They’re Building While these projects are grounded in distinct democratic contexts around the world, they are united by a shared hypothesis: that, if built differently, AI can strengthen democracy while preserving the human judgment it depends on.

Over the next 12 months, these ten project teams will advance their work through our Incubator, receiving coaching, expert advice, and peer mentorship as they build and test new ways to strengthen democracy in the age of AI.

_From Plural Reality (Japan):_ Inspired by democratic challenges in Otoineppu, a Japanese village facing population decline, this tool helps residents understand and express their views before participating in community deliberation. Uses an adaptive LLM questioning loop to clarify individual positions, then maps them with PCA-based opinion clustering to identify consensus, factual disputes, and value differences.

_From WeSolve Foundation (Philippines):_ Born from a grassroots anticorruption movement, this volunteer-run, open-source platform monitors how the government spends money, awards contracts, and manages public projects.

Uses a parsing-and-normalization pipeline that turns scattered government records into machine-readable open datasets, then layers automated red-flag analysis over them to surface the projects that journalists and citizens should scrutinize first.

_From Ciudadanía Inteligente (Chile):_ Piloted in Chile and built to expand across Latin America, this AI-powered early warning system tracks legislative activity and emerging threats to civil society. Ingests legislation and news through public APIs and classifies it with a LlamaIndex retrieval layer over a curated legal corpus, generating risk alerts that human analysts validate before they reach the organizations who act on them.

_(United States):_ Co-designed with city planners in New York City and organizers in North Carolina, this tool makes sense of large volumes of community input, surfaces recurring themes, and connects public decisions back to what residents actually said. Analyzes everything from sticky notes to recorded dialogue using RAG-grounded sensemaking, providing source-level traceability for every synthesized insight.

_From Civio (Spain):_ Grounded in more than a decade of defending the right to information in Spain, this tool helps people navigate the entire freedom-of-information process. Uses RAG to ground responses in Spain's transparency law, regulator resolutions, and court rulings, turning a plain question into a formal request, explaining the government's response, assessing whether a refusal is lawful, and drafting an appeal when it isn't.

_From CORRECTIV (Germany):_ Built for exiled journalists, researchers, and diaspora communities, this decentralized archive helps make invisible censorship visible by preserving censored journalism in Turkey, where tens of thousands of news articles have disappeared from public view since 2014.

Runs a smart-contract-coordinated miner network that monitors news outlets for deletions, then passes each through a verification pipeline before a human auditor signs off.

#### Open Facilitation Library _From Metagov (United States):_ As AI becomes more involved in public deliberation processes, this project works to define and teach what good facilitation looks like, helping train people and AI systems to support collective decision-making, self-governance, and meaningful public participation.

Uses two open-source elicitation agents: one using agentic RAG (Comhairle), and one that applies embedding-clustering (Harmonica). _From Dice Network (Czech Republic):_ Turning transparency obligations into actionable information about how European governments use AI, this platform transforms public records and freedom-of-information requests into a searchable public record of government AI use.

Combines a self-hosted registry, a freedom-of-information request engine, and an LLM document-analysis service to build searchable profiles of each government AI system.

_From Dean Issacharoff (United States):_ Developed to support the slow, fragmented work of accountability investigations, this tool helps investigators search thousands of hours of video to uncover evidence of human rights violations and other critical findings in minutes instead of weeks.

Builds a multimodal embedding index over large video collections, turning them into searchable databases that can be queried by what is seen or said on screen. _From Amplified Access (Uganda):_ Working across nine local languages in East Africa, this platform enables communities to report civic incidents through text, voice, and multimedia, helping local organizations identify emerging issues and respond more quickly.

Reports are processed through a speech-recognition and translation pipeline using fine-tuned, low-resource language models, then classified and plotted onto live maps. As we convene with experts and learn alongside our cohort, we’re hearing a lot of the same thing: **AI shouldn’t replace people or human decision-making.

It should instead help us become more informed, skilled, and engaged in the democratic processes that affect our lives. ** When it comes to AI and democracy, we’re not dealing with an optimization problem. We don’t need AI to automate away the essential messiness and friction of debate, disagreement, and compromise.

But that’s the real risk we’re facing: letting democratic life be treated as something to be streamlined, not strengthened. We don’t have to take that path. The future of AI is still ours to build.

AI can strengthen democracies when it’s designed to build understanding while keeping the messy but essential process of decision-making in our hands. This can take many forms. It can involve surfacing information that’s fragmented or difficult to access, supporting community discussions, building in transparency, and making participation more visible and connected to the decisions that shape our communities and societies.

Stay tuned as we gather more insights from our cohort, and watch our “AI for Democracy, or Democracy for AI? ” panel discussion to hear experts wrestle with some of the biggest questions defining the field today. Watch the panel discussion Watch the panel discussion * **Technology has always tested and shaped democracy.

** The printing press broke monopolies on information, fundamentally altering the architecture of power. The telegraph created high-speed networks that erased the barrier of distance, allowing discovery and dissent to travel across continents in an instant. Radio, television, and, eventually, the Internet, brought voices from the margins into the public square.

* **If history teaches us anything, it's that we can't afford to sit this moment out. ** Democratic progress is never a permanent win. It ebbs and flows: power recalibrates, adapting to new tools of influence.

Today we’re called to experiment, to create new spaces for imagination, test new ideas, and learn from one another. * **Because AI is already reshaping democratic life. ** We’re only just beginning to see how it’s reconfiguring systems and structures of power, yet the consequences are unfolding in real time.

The coalitions that sustain democratic life are becoming harder to hold together, placing even more pressure on democratic societies around the world. As power concentrates, we risk losing agency over the technologies that govern our lives and shape how we understand the world, and each other. And no single community or country has all the answers.

Progress depends on learning across borders and building on one another's experiments. * **The future of AI and democracy hasn’t been written yet. ** It’s still ours to build.

People around the world are already rejecting the status quo, adding to a growing counter-movement that turns imagination into infrastructure for experimentation, reinvention, and renewal. By choosing openness over secrecy and public benefit over narrow profit, we can create technologies that give power back to people and their communities.

Our Democracy x AI cohort is rooted in the belief that AI can still strengthen democratic life, but only if alternative technologies and ideas are supported early enough to survive, scale, and take root. By investing in people and projects that are building in the public interest, Mozilla Foundation is planting the seeds for a more open, democratic future with AI.

## Signals We’re Watching ### Core Areas of Democracy We received more than 1,000 proposals for the Democracy x AI cohort, offering early signals about how builders are thinking about AI and democratic life. Here’s what we’re seeing so far: * Our applicant pool was truly global, with 1,073 applications from 114 countries.

* Building Institutional Trust and Accountability emerged as the strongest area of interest, accounting for 44% of applications, followed by Defending Information Integrity (29%) and Protecting and Expanding Civil Space (28%). This pattern remained consistent from the initial proposal phase through the full application stage.

* Top-scoring applications were highly targeted and demonstrated a deep understanding of a specific community and problem, often within a clearly defined institutional, legislative, or civic context.

* Several areas attracted particularly high levels of interest and competition, including participatory governance tools, fact-checking and misinformation tools, legislative tracking and transparency, consensus-building and bridge platforms, and AI system auditing and governance.

Many of our finalist projects embraced practical experimentation, often grounding their work in distinct communities and political realities while testing ways AI might be used to address specific problems within democratic institutions, legislatures, and civic spaces.

### Learn More About Incubator Mozilla Foundation’s Incubator turns promising ideas into sustainable technologies that help build a better tech future: one that is more open, transparent, and designed to serve everyone. Discover what Incubator makes possible Discover what Incubator makes possible Something went wrong and your signup wasn't completed. Please try again later.

Stay updated with our newsletter Please enter a valid email address. - [x] I'm okay with Mozilla handling my info as explained in this Privacy Notice. Please check this box if you want to proceed.

If you haven’t previously confirmed your opt-in to a Mozilla Foundation-related email subscription you may have to do so now. **Please check your inbox or spam filter for an email from us to click and confirm your subscription**. * Annual Reports & Financials * Nothing Personal Magazine * Participation Guidelines Mozilla Foundation is a global non-profit and the philanthropic mission-holder of the Mozilla Project.

Most content available under a Creative Commons license. * [](https://bsky. app/profile/mozilla.

org) * [](https://www. instagram. com/mozilla/) * [](https://www.

linkedin. com/company/mozilla-corporation/) * [](https://open. spotify.

com/show/0vT7LJMeVDxyQ2ZamHKu08) * [](https://www. tiktok. com/@mozilla)