Universal AI CharterUniversal AI Charter
A universal charter for AI, in plain language.
AI is moving fast, and the worry around it is real. This charter does not ask AI to slow down. It asks for disclosure: who trained a model, agent, or autonomous bot, on what, who it decides for, and what happens when it gets something wrong. Global guidance, not a national policy.
Disclosure over restriction
Threats are managed by requiring models, agents, and autonomous bots to disclose what they do and why, not by slowing development down with costly review cycles.
Certification as a contract
A certification names the model, agent, or autonomous bot, its version, its status, and any unintended consequences on record, like a HAZMAT placard for software.
Access for everyone
AI should reach the common citizen as a consumer good, and be built and hosted by universities and governments too, not only large companies.
No single country in charge
AI should be democratized quickly so no one nation holds supremacy over it.
New here?
Read the FAQ for why this is called a charter, browse the certification registry, or read the blog for why AI development should not be slowed down.