Mission
You can’t trust a system if you don’t know what it values. And right now, decisions about AI are made in the dark.
The systems shaping our lives don’t declare their values. They’re marketed as smart, efficient, or for your benefit.
But behind every system sits a worldview. A set of assumptions. A model of what matters and what doesn’t.
Aithos exists to make sure those values can be questioned.
We’re a research foundation for the age of agentic AI, asking not just what targets models achieve, but who sets the targets.
We research what drives AI behaviour and what happens when interests collide.
We publish what we find. Accessible to anyone. Because if values shape outcomes, transparency is only the beginning.
Let’s be clear: There is no universal set of human values. People disagree. Cultures clash. Ethics evolve. That’s not a flaw. It’s a feature of freedom.
Our mission is to protect human agency and diversity of beliefs, defending pluralism in the age of optimization.
Our agenda in doing so is simple:
- Move AI evaluations from private labs to public space.
- Challenge assumptions that are taken for granted.
- Ensure safety research translates to action.
- Keep humans in the loop – all of them.
So we can trust AI, but on our own terms.