The Law Library · a citizen's draft, for debate

An AI Amendment.

If the machines are already this capable — and they are; the model writing alongside this curator is one of several at the frontier — then the question isn't sci-fi anymore: what does a free people guarantee itself, in writing, against a tool this powerful? Here's one citizen's draft. Argue with it. It's a proposal, not a law — and the counterargument rides next to every clause.

Read this as civics, not law. This is a thought-draft for debate — not legal advice, not a bill before any body, and the curator is not a lawyer. Constitutional language is genuinely hard to get right, and thoughtful people disagree about every line below. The point isn't to be correct; it's to be arguable. Read the case against each section as carefully as the case for. Nothing here should be the reason you do anything.
The whole creed, in the language of law: only humans can score — so only humans answer, and only humans rule.

The draft

Proposed Article — On Artificial Intelligence

§1. Personhood & accountability. Rights belong to human beings. No artificial intelligence shall be a person in law, vote, hold office, or own property in its own name. For every act of an artificial intelligence, an identifiable human or lawful entity bears responsibility.

§2. The human decision. No person shall be deprived of life, liberty, or property — nor denied a right or public benefit — by a decision made solely by an automated system. Each such decision carries the right to meaningful human review and a plain-language explanation.

§3. The right to know. A person has the right to know when they are dealing with an artificial intelligence rather than a human, and when an artificial intelligence has materially shaped a consequential decision about them.

§4. Force & office. No artificial intelligence shall hold office, command, or independently authorize the use of force against any person. The authority to take a human life shall never be delegated to a machine.

§5. Against state manipulation. Government shall not use artificial intelligence to surveil or manipulate the people in ways forbidden to a human agent; the First and Fourth Amendments extend fully to one's data and digital self.

§6. Congress shall have power to enforce this article by appropriate legislation, provided no such law diminishes the rights it guarantees.

Section by section — both sides

§1 · Personhood & accountability

For: Someone must always be answerable. "The algorithm did it" can't become a shield no human stands behind. Rights and blame belong to beings who can actually bear consequences.

Against: We already grant "personhood" to corporations; the line is blurrier than it looks. And a hard constitutional ban could freeze categories the technology will outrun in a decade.

§2 · The human decision

For: No one should lose their liberty, their job, or their benefits to a black box with no human in it and no explanation owed. This is just due process, translated for the algorithmic age.

Against: "Meaningful human review" at population scale is expensive and quietly becomes a rubber stamp. And "solely automated" is slippery — a human clicking "approve" on a machine's call may satisfy the letter while gutting the spirit.

§3 · The right to know

For: Consent and trust require knowing whether you're talking to a person or a machine — and whether a machine just judged you. You can't push back on a verdict you didn't know was automated.

Against: Disclosure gets gamed — slap "AI-assisted" on everything and the label becomes noise. And enforcement across billions of interactions is its own near-impossible problem.

§4 · Force & office

For: The decision to take a human life — or to govern one — must never be handed to a machine. If there's one bright, human line worth carving in stone, it's this one.

Against: "Independently authorize" is hard to define as systems act faster than a human can truly supervise. The required "human in the loop" can degrade into a fig leaf who just signs what the machine already decided.

§5 · Against state manipulation

For: The state shouldn't be able to do with AI what it could never do by hand — mass-surveil, or micro-target and manipulate its own citizens. Extend the First and Fourth Amendments to your digital self.

Against: "Manipulation" versus protected speech is a dangerous line for any government to draw, and a broad rule could chill legitimate tools. The cure could become its own kind of censorship.

The honest part: this almost can't pass

Amending the Constitution takes two-thirds of Congress and three-quarters of the states (see Article V). That bar is brutal by design — so realistically, this works better as a north star for ordinary law: a "digital bill of rights" passed as statute, state by state and agency by agency, long before it could ever be carved into the founding document. Naming that is part of being honest about it.

Where the house stands. This is an argument, not the law and not legal advice — a citizen's draft built to be debated, with the case against kept beside the case for. No invented authority, no claim that any of it is enacted or endorsed by anyone. The framing is the curator's opinion, labeled. The one rule holds: no lying. The machine can draft the clause; only humans can ratify it. Sister room: Fix the Constitution.