The ethical principles that guide every tool we build and every decision we make.
These values aren't marketing copy—they're commitments we code into our tools. Informed by frameworks for ethical AI from organizations like Neighborhood AI and research institutions worldwide, these principles shape how we build technology for communities.
The AI industry's environmental footprint is enormous and growing. Training large language models consumes massive amounts of energy, and every inference request to frontier models adds to global carbon emissions. We believe that building civic technology should not come at the cost of our planet.
Our commitment to climate justice means prioritizing locally-run LLMs over frontier models wherever possible. Local models use a fraction of the energy, don't require data transmission to distant data centers, and can run on existing community hardware.
This isn't just about reducing our carbon footprint—it's about modeling a sustainable path forward for AI development that other civic technology projects can follow.
We prioritize models that run on community hardware rather than corporate clouds.
We use the smallest model capable of the task, not the most powerful available.
Our tools are optimized to minimize computational overhead.
We're honest about when we still rely on frontier models and our timeline for transition.
We believe that civic technology should be as transparent as the democracy it serves. Every line of code we write is open source, available for anyone to inspect, audit, critique, or improve. There are no black boxes in our stack.
This radical openness isn't just about accountability—it's about building trust. When community members can see exactly how their data is being processed, they can make informed decisions about whether to use our tools. When developers can audit our code, they can identify vulnerabilities we might have missed.
Open source also means our tools can outlive any single organization. If The Community AI Project ceased to exist tomorrow, the code would remain available for communities to maintain, fork, and evolve on their own terms.
All our code is published under permissive licenses.
We explain not just what our code does, but why we made the choices we did.
We actively welcome and incorporate feedback from the communities we serve.
Our tools are designed to be portable and self-hostable.
AI should amplify human connection and civic action, not replace it. Our tools exist to reduce the tedious administrative friction that keeps people from engaging with their communities—not to automate away the human elements of democracy.
This means we're deeply skeptical of AI solutions that claim to "replace" human judgment in civic contexts. Government meetings need human witnesses. Community organizing needs human relationships. Democratic deliberation needs human voices.
Our tools help people find relevant information faster, translate complex documents into accessible language, and discover connections they might have missed—but the decisions, the actions, and the relationships remain human.
We help humans do their work better, not replace them.
Critical decisions always require human review and approval.
Our tools facilitate human connection, not substitute for it.
Users remain in control of their data and their choices.
We believe neighborhoods should own their digital infrastructure just as they own their physical spaces. The data generated by communities should serve communities, not be extracted for corporate profit.
This principle shapes everything from our data architecture (local-first, minimal collection) to our governance model (community input on feature priorities) to our business model (no advertising, no data sales, ever).
Inspired by the vision of Neighborhood AI for "hyperlocal, community-governed AI for dignified civic infrastructure," we're building tools that can be genuinely owned and operated by the communities they serve.
Community data stays in the community.
We never sell data or serve advertising.
Communities can run our tools on their own infrastructure.
Feature priorities are shaped by community input.
Our values are shaped in collaboration with Neighborhood AI, working toward hyperlocal, community-governed AI infrastructure.
We design for disabilities and language barriers first, ensuring no neighbor is left behind. Accessibility isn't an afterthought or a compliance checkbox—it's a core design principle that shapes our tools from the ground up.
This means screen reader compatibility, keyboard navigation, high contrast modes, and clear visual hierarchies. It also means plain language alternatives to jargon-heavy government documents and translation support for the languages actually spoken in our communities.
Truly accessible civic technology opens democracy to people who have historically been excluded—not just technically, but practically. If someone can't use a tool because of how it's designed, then that tool has failed its civic mission.
We follow web accessibility guidelines as a baseline, not a ceiling.
Tools work in the languages communities actually speak.
We translate government jargon into accessible explanations.
Tools work on slow connections and older devices.
We build lightweight, resilient tools that run on old hardware and don't require expensive upgrades. Sustainable technology isn't just about environmental impact—it's about building tools that communities can actually maintain.
The tech industry's obsession with the latest and greatest creates digital divides. When software requires the newest hardware to function, it excludes people who can't afford constant upgrades. When tools depend on complex infrastructure, they become impossible for small communities to self-host.
Our tools are designed to be simple enough that a dedicated community member can understand and maintain them, performant enough to run on modest hardware, and robust enough to keep working when infrastructure fails.
Fewer moving parts means fewer things that can break.
New features don't break existing workflows.
Tools remain useful even when some features aren't available.
Every tool is documented well enough for community maintenance.
These aren't abstract principles—they're embedded in every tool we build. See them in action.