If philanthropy wants to push AI in a better direction, now is the time.

Today, Mozilla is publishing its AI Funding Principles — institutional lessons learned from five years and $75 million of trustworthy AI grantmaking around the globe.

While we had long planned to publish these at the close of 2023, releasing them at this moment now feels serendipitous. Discussions and debates about AI — how to build it, how to regulate it, how to fund it — have been mounting all year. In the past few weeks, they’ve reached a crescendo: the AI Safety Summit, the White House’s encouraging executive order, and a $200 million philanthropic commitment from Mozilla and our allies.

At this crucial moment, these principles can help inform like-minded philanthropists, civic justice organizations, activists, and technologists — people who believe the dominant AI paradigm should be one of individual agency and corporate accountability.

Five years ago, Mozilla spotted a problem that catalyzed our decision to focus Mozilla’s grantmaking on AI. AI was emerging as (and now is) the next phase of computing. It’s at the very core of many new technologies and capabilities. And it’s used to power everything from social media to voice technology to the digital machinery of government.

At the same time, AI was centralized — dangerously so. And still is. Despite its influence over billions of lives, it's a technology controlled by just a handful of corporations with, usually, a singular incentive: profit. This has all sorts of problems: for equity, for privacy, for agency, for accountability.

About five years ago, we sought to point all of Mozilla’s philanthropic and movement-building might at this problem, by shifting power to communities building alternatives. Since then, we’ve provided over $20 million in grants across five continents: Africa, Asia, Europe, North America, and South America.

These principles can help inform like-minded philanthropists, civic justice organizations, activists, and technologists — people who believe the dominant AI paradigm should be one of individual agency and corporate accountability.

J. Bob Alotta, Senior Vice President of Global Programs at Mozilla

The solutions we’ve funded have been diverse: agriculture-focused chatbots for farmers in Tanzania. Apps to mitigate online gender violence in India. Gig worker cooperatives in the UK. Tech policy fellowships in Brazil and the U.S. But they all have a common thread: making AI more trustworthy.

These Funding Principles sum up what we’ve learned along the way. In open-source Mozilla fashion, we’re sharing them so others can benefit from them, build on them, add to them.

There are 10 total. They address everything from the importance of context (“Be geographically specific”) to who the true protagonists should be (“Acknowledge that expertise lives at the site of the experience.”). And each one features case studies from Mozilla’s own grantmaking: Common voice, MozFest, fellowships and awards, and more.

This is our invitation to the philanthropic community to join us and double down on thoughtful, impactful grantmaking around trustworthy AI. Together, we have a chance to push AI in a better direction.