Nabiha Syed: “Philanthropy can fund the AI people need before the market sees its value”

Courtesy of Nabiha SyedFounded in 2003, the Mozilla Foundation works towards an open, accessible digital sector that serves people. In particular, it funds and builds AI that puts people in control. An interview with Nabiha Syed, executive director of the Mozilla Foundation, who calls on philanthropy to support alternatives to the dominant models.

What is ‘trustworthy’ artificial intelligence?

Technology should be worthy of our time and resources. For the Mozilla Foundation, trustworthy AI must enable people to understand how the system works, to opt out of it, to challenge its decisions and to keep control of their data. Can you leave, take your data with you, and choose something else? Choice requires alternatives. We therefore need to look beyond technical questions alone and ask: who owns the data? Who sets the defaults? Who can examine how the system works? These decisions cannot be left solely to the companies that profit from them.

But we also ask other questions: is this technology worth our time? Does it meet the needs of people and the planet? Are the resources it draws on worth the cost? Does it improve people's quality of life or enrich a few? The value of a technology cannot be defined by a single company or a single government. It must be discussed with the people directly affected. Philanthropy can help people shape those decisions, and support the alternatives that make their choices real.

What major risks do you see in the development of AI, at a time when power and resources are concentrated in the hands of a few large tech players?

The main risk is a ‘failure of imagination’. Concentrated power becomes concentrated imagination: a few companies get to decide what technology is for. Today, a few companies control much of the infrastructure, capital and distribution, and so they can set the terms of access to technologies. In doing so, they define the realm of the possible before society has even been able to decide what it wants to do with these technologies.

This concentration of resources also carries very concrete risks: discriminatory bias through biased data, cybersecurity vulnerabilities and increased dependence on a limited number of technological infrastructures. That is why much of our work involves building infrastructure - like incubating and spinning out Mozilla Data Collective - that enables those who want to develop alternative technologies to actually do so.

Imagination at the application level is also critical. For example, we support Adalat AI, a non-profit that develops AI tools to reduce court delays, in particular through the automated transcription of hearings. The aim is to show that there are other ways of developing AI, suited to the needs of society and of people. We call this resistance through reimagination: challenge the systems that fail people, and give builders the means to create alternatives.

What role can open source play in developing AI that respects people and their freedoms?

Open source lets people inspect, adapt and build on technology, rather than simply accept someone else’s terms. Openness makes scrutiny possible. It still takes people, resources and accountability to make that scrutiny count. Too often, particularly in the United States, safety and freedom are presented as opposing notions. Yet safety can be precisely what gives us more freedom: we are freer when we can use a technology without fearing that our data will be collected or passed on without our knowledge. Open source is not the answer to everything, but it helps to foster the development of alternative technologies: testing systems, evaluating them, identifying biases, comparing their performance.

‘I agree’ is a click. Control is an ongoing say in how your data is used. Open source allows for control, not an inconvenient moment that looks like consent. People must be able to say: I don't want you to collect this; I don't want it used in schools; if you use it, I want to be compensated; if you have used it in a way I don't accept, delete it. Real data control covers the whole chain: collection, access, reuse, compensation, deletion and governance.

Open source also rests on a form of humility: if we all take part, other people can identify risks that we do not see. Philanthropy can fund the maintainers, independent testing and community governance that openness needs to work.  

What role can philanthropy play in the development of AI, given the considerable means of tech companies?

Philanthropy can be first-believer capital: supporting useful technology before its value is obvious, and helping it survive long enough to prove itself. We will never have the means of the commercial AI players. Philanthropy, however, can take risks that they do not and ask other questions. It can create spaces in which to experiment with and imagine public interest AI, without the perverse incentives of rapid financial returns, by stepping in early and committing for the long term.

We also need to challenge the notion that innovation comes first from tech companies, with philanthropy coming in only afterwards. We can ask different questions from the very outset: have the people affected been consulted? Has the technology been tested? Does it protect human dignity? What might its long-term consequences be? The changes that social media has brought about in our attention and our relationships with others should prompt us to start anticipating the effects of AI now.

The arts and culture also have an essential role to play in thinking these questions through. Artists, filmmakers and writers help us ask what technology is for, and build things engineers might never think to make. They are skilled at observing trends and imagining their consequences. Philanthropy can give them the means to explore these issues and in doing so feed our capacity to anticipate. The same is true of academic research and independent journalism, which produce indispensable knowledge. We must preserve their capacity to imagine other futures.

You also advocate closer cooperation between foundations. Why?

Somewhere in the world, someone is probably building the technology we actually need, not simply the one others are trying to sell us. The role of philanthropy is to spot these initiatives very early, to take a risk on them and to help them grow.

But each philanthropic organisation spots different initiatives. A cultural foundation will not focus on the same things as a foundation specialising in technological innovation or social justice. Working together makes it possible to combine these perspectives and to identify under-resourced areas. Shared intelligence helps us find these builders. Pooled funding lets us make a meaningful commitment and stay with them. Mozilla Foundation is investing in both and eager to partner with Europe.

Coordination can also be a way of making demands collectively. Imagine that a coalition of foundations decided to support only projects using AI systems tested for bias. It would send the market a clear signal about the standards expected.

How can citizens be given more power in the face of AI?

People need the power to shape technology, refuse it, and choose an alternative. First, we must remind citizens that they have a choice, including the choice not to use a technology. As part of Creative Futures, a Mozilla Foundation initiative that brings together artists and technologists to imagine other uses of digital technology, we asked artists around the world how they would like to use AI. Many told us they would like to use it for the tasks they dislike, such as bookkeeping, but not to create their art. We distilled these expectations into the report 8 Rules for AI from Hollywood to Silicon Valley, which went on to feed into legislative discussions in California.

Philanthropy can play this role: going out to meet the people affected, bringing their expectations to light and carrying them into public debate. Listening is the starting point. Power means those demands change what gets funded, built and deployed.

What are the links and complementarities between the American and European philanthropic approaches?

The opportunity is to connect philanthropic capital across borders with builders who understand local needs, and give those builders access to shared infrastructure. In the United States, the tech ecosystem is operating within constraints that have a set of ideologies, which then seek to spread. That means that existing discrimination will get reproduced. In Europe, as in other regions of the world that are not subject to the same political environment, it is possible to build differently, to test systems, to examine their risks and to innovate around their safety. The European approach is extremely valuable because it fosters alternatives.

The question of sovereignty also looms large in Europe. If we do not want to depend solely on the United States, we must build other infrastructure. It was with this in mind that, this year, we incubated Mozilla Data Collective, built by the team behind Common Voice. Launched in 2025, it became a standalone, mission-locked British company in 2026. By May 2026, more than 190 organizations were sharing over 600 datasets across more than 300 languages. Mozilla Data Collective enables this data to be created, selected and controlled on terms chosen by those who contribute it. This initiative shows how philanthropy can help open infrastructure to emerge. Sovereignty becomes real when people have infrastructure they can build on and alternatives they can choose.

I also find Europe’s approach to questions of safety and human flourishing inspiring. In the United States, the first question asked is often ‘Is this innovation possible?’, whereas in Europe it tends to be ‘Should we develop it, and what will its impact be on human beings?’ The question is not only what technology can do, but what human beings can do with it. This is why we are trying to popularize the idea of product-community fit. Alongside product-market fit, we ask about product-community fit: does it solve a problem the community recognizes, on terms it can shape? Communities should decide how AI should be apply to their education, their healthcare, their creativity, or anything else that you use technology for. We judge success by whether communities use it, benefit from it and retain meaningful control. Different cultural and geographical contexts can give rise to many forms of innovation. It is essential for philanthropy to support this diversity.