Why Curos chose to be a non-profit

When I explain to people that Curos is a non-profit AI organization, the reaction is usually polite confusion. “Non-profit” reads as “charity,” and AI doesn’t look like charity. It looks like the most expensive technology ever built, which is true, and it looks like a market, which is also true. So why would anyone build it outside the market?

I want to give the real answer, because it is the reason this organization exists at all.

AI is becoming infrastructure

Consider what is happening to everyday life. People get their questions answered, their writing helped, their work drafted, their decisions informed, their children educated — increasingly by models. AI is not a feature of the economy anymore; it is becoming a layer of it. Like roads, like the power grid, like search engines before it, it is infrastructure.

We hold infrastructure to a different standard than products. We do not accept that the road system answers to whoever pays the most. We do not accept that the power grid turns off for people who cannot afford it. We built agencies, regulations, and public ownership precisely because infrastructure is too important to be governed by the short-term interests of its owners.

AI is not there yet. It is still treated as a product, answerable to its shareholders and priced for those who can pay.

The structure decides the outcome

Here is the thing I have come to believe after years in the field: the structure of the organization decides the outcome of the technology, not the intentions of the people in it. A company is legally obligated to maximize returns for its shareholders. That obligation shapes what gets built, what gets priced, what gets published, and what gets cut. No individual in that company is a villain; they are all doing exactly what the structure asks.

So the question is not whether we can find good actors inside companies. It is whether the structure itself serves the public.

What a non-profit changes

Building Curos as a non-profit changes specific, concrete things:

• No shareholders to maximize for. Every decision is measured against the mission: does this serve the people who use it? A product decision that helps users but shrinks the bottom line is a win, not a loss.
• No sale. The organization can never be acquired, because there is no equity to acquire. Whatever we build stays ours, in the non-profit sense — which means it stays available to the public.
• No data to monetize. We do not sell ads, and we do not sell data, because there is no reason to. The product’s incentives are aligned with the user’s interests, not with an advertising revenue line.
• Accountability as a feature. Non-profits are regulated, audited, and — if they want to survive — transparent. We want to be held to that standard.

The honest caveat

Non-profit does not mean good. There are non-profits that fail their missions, and there will be times Curos falls short of its own commitments. Structure is a constraint, not a guarantee.

But constraints matter. If you want AI to serve everyone, you cannot start with an organization legally obligated to serve its owners. You start with an organization that has no owners — only a mission.

That is what we chose. It is more work, it is slower, and it is less certain than raising venture capital would have been. But the technology is too important to bet the other way.

All news →