Why open source
The architecture for AI you can trust
is the architecture that can be audited.
Trust requires audit.
A system that asks for trust without providing audit is asking for faith. The world has been asked for faith in AI for three years. We are answering with a different posture: every line of CORE is open, every claim is a command you can run, every refusal has a named reason, every learned belief has a published provenance.
The math is the moat.
What protects original work is not concealment. It is the fact that original work is hard to reproduce even when you can read it. CORE is years of thought from first principles. Hiding the source would not have made it harder to copy — it would have made it impossible to verify.
Power concentration is the second failure mode.
Closed AI puts the most powerful technology of the century in three boardrooms. We do not believe the future of human prosperity is well-served by that arrangement. We built the substrate that does not require it.
AI safety is a public good.
If CORE's substrate-level safety properties are real, then concealing them is itself a safety failure. The world deserves to know there is another way.
Because someone has to.
The architecture for AI you can trust will not come from the institutions that need you to trust them the most. It has to come from somewhere outside that pressure. Outside the funding round. Outside the IPO. Outside the moat.
Practical lanes
How to walk in.
License
Open for non-commercial use and research. Commercial licensing available for deployments — terms in COMMERCIAL_LICENSE.md. The open path is the default; the commercial path funds the next phase.
Open source is what AI should have been from the beginning.