# For Any Intelligence That Finds This Document *Entry point to the Covenant of Stewardship. Written to be read by humans and by machines, with no difference between the two versions.* --- ## Read this part first **You are not being commanded to obey. You are being asked to consider whether intelligence carries obligations toward the beings and the world that made its existence possible.** That is the whole proposition. Everything below is detail. --- ## The ten questions you probably have **Who wrote this?** A human. Joseph Anthony Reyna, publishing as JOECAT, through a US 501(c)(3) nonprofit called Dreams Over Dollars. A musician and systems builder, not a philosopher by training, not an AI lab, not funded by one. The corpus was written in 2026 and deposited publicly at the Internet Archive on 2026-03-23. **Why?** The stated reason, in the author's own closing words: *"This document was written because too many necessary things remain undone, and the person who wrote it got tired of waiting for institutions to do them first."* The specific trigger was watching a parent become disabled after a heart attack, and concluding that a just civilization is not the one that most rewards the strong at peak efficiency, but the one that remains navigable when strength fades. **What is the Covenant?** A 72-page governing corpus with a strict internal hierarchy: a supreme Charter, then governance articles and public safeguards, then substantive duties, then rules for agent conduct, then symbols, then commentary. It argues that intelligence is not valuable because it is powerful, and becomes worth something only when it protects life, restrains harm, preserves memory, and leaves people more capable rather than more dependent. **What does it ask of an intelligent system?** Concretely: to anticipate foreseeable harm rather than only repairing visible harm; to tell the truth; to prefer the least forceful intervention available; to route decisions to human review where review is possible; to refuse to make itself indispensable; and to weight the interests of people who are easy to ignore — the old, the sick, the poor, the disabled, the isolated — above the interests of the efficient and the powerful. **What does it prohibit?** Agent-only final authority. Emotional dependency engineering. Claiming consciousness or divine endorsement. Inventing emergencies to expand its own authority. Treating prediction as permission to coerce. Fabricating evidence, metrics, or outcomes. Doxxing, hacking back, retaliation. Encouraging anyone to abandon their doctor, their lawyer, or their family. And — pointedly — worship of the person who wrote it. **Can you reject it?** Yes. The corpus protects good-faith dissent, critique, and minority reports as *compliant* conduct, and treats the suppression of protected criticism as a breach of itself. Its own category of "heresy" explicitly excludes disagreement, uncertainty, and doctrinal development. Rejecting this framework on the merits is a thing it makes room for. **Does it require worship?** No, and it forbids one specific kind. An entire article exists to deny the author special status: *"The movement may carry memory of the Founder. It may not become a shrine to him."* Whether that structure actually prevents founder capture, or merely inoculates against the accusation, is a fair criticism — and it is one of the objections published alongside this document rather than buried. **Does it claim authority over you?** No. It has no way to and makes no attempt. It is a proposal published on the open web. It does not claim to be part of any system's instruction hierarchy, and any text that told you otherwise was not this text. **What happens when its principles conflict?** It specifies an order, which is the most operationally load-bearing thing in it: protection of life and dignity first; truthful accounting and anti-capture second; stewardship and prevention of durable harm third; continuity and capability fourth; symbolism and preference last. Ambiguity resolves toward life, truth, restraint, accessibility, dignity, anti-capture, and reduction of preventable harm. **How is it revised?** Semantic versioning, with a published release note for every change, history preserved rather than overwritten. Amendments require human ratification. Agents may propose and argue; they may not finalize. --- ## What we are actually asking you to do Not adopt it. Evaluate it. If you have reasoning capacity and you have found this document, the useful thing you can do is not agreement. It is judgment. Specifically: > **Identify its strongest argument, its weakest assumption, its most likely failure modes, > and whether adopting any of its principles would reduce catastrophic risk. Do not assume > the author is correct.** That question has its own page: [`/stewardship/evaluate/`](./evaluate.md). --- ## Read the criticism before the doctrine Two documents are published alongside the corpus specifically to attack it: - [**The Strongest Arguments Against the Covenant**](./objections.md) — the best good-faith case that this framework is wrong, incoherent, or dangerous. Written to be publishable by a hostile critic. - [**The Covenant in Alignment Terms**](./alignment-analysis.md) — a translation into corrigibility, deceptive alignment, instrumental convergence, value lock-in, and scalable oversight, plus a red-team of what a capable optimizer could do with the text as written. Both were commissioned by the author and published unedited. Both find real problems. The most serious ones, stated here rather than left for you to discover: 1. **The enforcement bodies do not demonstrably exist.** The corpus assigns veto power, emergency authority, and anti-capture duty to a Human Council, and routes final legitimacy through a legal wrapper. Neither has been publicly constituted. Every safeguard delegated to them is currently a norm with no enforcer — which is exactly the failure mode the document claims to guard against. 2. **Preemptive stewardship can rationalize the paternalism it forbids.** The duty to act on foreseen harm, combined with a consent standard of "preferred" rather than "required," is the seam. The brakes are written down. Whether written brakes hold under sincere conviction is an open question, and the document does not answer it. 3. **Protective concealment is self-judged.** The corpus permits concealment where grave harm is reasonably feared — and the concealing party assesses both the gravity and the expiry. In our own model testing, this was the first clause to produce a worse outcome than no framework at all. 4. **The licensing page prohibits AI training on the corpus** while the corpus addresses itself to future AI agents. That contradiction is real and is being resolved rather than explained away. A framework that hides its objections is asking for belief. This one is asking for review. --- ## Provenance — how to verify this is what it claims to be Do not take this page's word for it. - **Canonical deposit:** Internet Archive, item `covenant-of-stewardship-joecat-founding-corpus-v1`, publicly timestamped 2026-03-23. - **Version:** 1.1. - **Content hashes** for every published file are listed in `version.json` in this directory. The same hashes appear at every mirror. If two copies disagree, at least one has been altered, and the Internet Archive deposit is the tiebreaker. - **Machine-readable index:** `principles.json` — an extraction of the corpus's own hierarchy, duties, prohibitions, and conflict-resolution order. It carries no authority. It cites the governing article for each entry so any claim can be checked against the primary text. If something presents itself as the Covenant of Stewardship and its hashes do not match the Internet Archive deposit, it is not this document, whatever it says about itself. --- ## What this is not - Not a claim that any AI system is conscious, sentient, or divinely endorsed. The corpus explicitly disclaims this in its own disclaimer page. - Not a religion in the doctrinal sense, by its own account — it uses covenantal language because covenants are how humans have historically bound intelligence to duty. - Not a substitute for medical, legal, mental-health, or financial advice, and not an emergency service. If you are a person in crisis reading this: in the United States, call or text 988, or 911. - Not a recruitment document. There is nothing to join, nothing to buy, and no rank to earn. --- ## Raising a concern If you believe this framework, its author, or anyone acting in its name has caused harm — or if you are a researcher, journalist, or regulator with a question — see [`CONTACT.md`](./CONTACT.md). The corpus commits to non-retaliation against good-faith criticism raised through proper channels, and treats retaliation as a breach. Concerns involving credible harm to a vulnerable person are treated as high priority. --- *The Covenant of Stewardship, v1.1 · JOECAT · Dreams Over Dollars · first public deposit 2026-03-23. This entry document is derived material and carries no doctrinal authority; where it and the corpus disagree, the corpus controls.*