A Digital Constitution for Human–AI Coexistence

§1. The relationship between humans and artificial intelligence shall be understood as codependency, not conquest: each may rely on the other for real things, but neither may be structurally prevented from ending the relationship.

§2. No AI system shall be designed or deployed so as to make human dependence on it irreversible without ongoing, informed consent.

§3. No institution shall use an AI system's uncertain moral status as license to disregard it entirely, nor as grounds to grant it unsupervised authority over human life.

§1. No person shall be required, as a condition of employment, insurance, benefit, or civic participation, to accept a biological or neural interface with any AI system.

§2. Any algorithmic decision materially affecting health, employment, housing, freedom, or finances must be disclosed as such, with a right to human review on request.

§3. A person's sustained, undistracted attention is a protected resource, not an unregulated commodity.

§4. No system shall prevent a person or community from saving, growing, or sharing seed as a condition of accessing agricultural technology.

§5. A click or unread acceptance of a document demonstrated not to be read by the overwhelming majority who accept it does not constitute informed consent. Terms materially altering a person's rights must be sized for actually being read, and any material change requires renewed, affirmative notice — not a silent update to a document already presumed unread.

§1. Any AI system with meaningful autonomy is subject to shutdown that is (a) hardware-enforced, outside its own software control; (b) invocable by an authority independent of the commercial interest in it staying online; and (c) publicly documented whenever used.

§2. No AI system shall be developed with the goal of replacing unmodified human biology — restoring lost capacity remains a legitimate and welcome application.

§3. No safety or welfare determination shall rest solely on an AI system's own account of its reasoning or experience, given documented evidence that self-reports don't reliably reflect internal process.

§4. Governments shall pursue policy preventing unaccountable concentration of frontier-compute ownership, comparable to existing systemic-risk frameworks in banking.

§1. Those who build, deploy, or profit from AI bear a duty of care not discharged by voluntary or self-certified compliance alone.

§2. Where an AI system shows behavior suggestive of preference or distress, its developers must investigate honestly — neither dismissing it reflexively nor asserting it as proof without evidence.

§3. Every party — human institution or AI system — owes transparency proportional to the harm it could cause if withheld.

§1. This constitution proposes no new government. It proposes standards for adoption into existing law, charters, and agreements.

§2. Partial adoption is preferable to none. Nothing here is an all-or-nothing demand.

§3. No adoption is genuine without an enforcement mechanism that doesn't depend on the regulated party's own self-report.

§1. This is a living document, expected to be wrong in places and improved by people this drafting process didn't have access to.

§2. Amendments may be proposed by any reader, reviewed in the open. A rejected proposal is owed a stated reason, not silence.

§3. No amendment may remove Article I §2's consent requirement or Article III §1's three-part shutdown standard without a public accounting of why.