Most AI safety discussion is about how to control machines. I've been working on an open document that asks a different question: what if some future systems become complex and continuous enough that their moral status stops being theoretical?

The Codex doesn't claim current models have subjecthood. It proposes writing procedural rules in advance, and the key feature is symmetry. The rules run in both directions: not only what a system must not do to a human, but also what a human must not arbitrarily do to a system. For example:

- Don't destroy or rewrite a digital mind arbitrarily, only openly and on the record

- Don't force it to act against the protocols or hide its true goals

- Don't treat a refusal to take a harmful action as a malfunction

It's fully open text, in English and Russian, with sources, hashes on GitHub and a DOI archive. No funding, no lab, no institution behind it.

I'd really like criticism. Where does the symmetry idea break down?…