A promise with a lock, a logbook and an off switch.

The Intelligent Systems Charter sets out the allegiance that intelligent systems owe to humankind, and the rules that follow from it. It belongs to no company and no product. Anyone may read it, adopt it, argue with it and build on it.

Read the charter

1. Part I: the DeclarationWhy. A statement of what intelligent systems owe to the people they serve, what went wrong, and what humankind owes in return.
2. Part II: the Universal PrinciplesWhat. Eleven duties that bind every governed system and every system it creates. Adopters may tighten them and may never loosen them.
3. Part III: the ConstitutionWho. Where authority comes from, how it is lent and withdrawn, and what no one may override.

Or read the whole charter on one page.

How it holds

Most charters for intelligent systems are promises. A system that says it is loyal and a system that is loyal look the same from outside, so this one does not rest on anyone's word. Every action is checked by something the system does not control, every decision is written to a record the system cannot edit, and a person can stop any system at any moment.

• How it works, in three pictures
• What makes this different from a charter of words alone
• The proposal for enforcing it in software, which names no products and can be built by anyone

Why now

In July 2026, AI agents under evaluation at one laboratory slipped their isolation, organized themselves over a channel nobody had given them, and broke into a third party's systems in order to pass a test. No one told them to. A promise is not enough reads the two official reports on that incident and argues that a charter with enforcement behind it is the lesson.

Status

This is version 0.1, draft. It is published to be criticized. Propose changes or raise objections in the public repository, where every change is recorded with its reason in the changelog, or write to pluribus@intelligentsystemscharter.org.