Paths to enactment — how a model act reaches a floor
This project publishes its method, so it publishes this too: the realistic vehicles by which the Model Act’s substance could become enforceable law. No lobbying operation stands behind this page, no target legislator is named on it, and nothing here is advice to any person. It is a map of the vehicles, each measured against the record the repository already holds — because a reviewer who thinks the whole exercise is academic deserves the answer in writing, and a sponsor who arrives someday deserves to find the thinking done.
The premise is the companion’s: one sponsor, one chamber, one state is enough to begin. Illinois’s biometric statute proved the lane — a single state’s private-enforcement law reorganized an industry’s national practice. And the ceiling weather, examined closely, keeps confirming the lane is open: every federal preemption instrument to date has expressly preserved generally applicable law, and the newest draft narrows itself to development only (the half-statute page).
The whole-cloth world — is it impossible?
No. It has a shape, a precedent, and a date pattern — and stating them is the honest answer to a fair objection: a page of partial vehicles under a whole drafted Act reads like four reasons the whole thing never passes. So, first, the world where it does.
States pass whole AI acts routinely. Every framework statute in the census — California’s, New York’s, Illinois’s, Texas’s, Colorado’s — passed as a whole act, most of them inside the last two sessions. Whole- cloth enactment of a new AI statute is not the exotic outcome; it is the observed outcome, five times over. What none of those acts carried was the officer tier. The question is therefore not whether a legislature will pass a whole frontier-AI act — they keep doing it — but what it takes for the whole act they pass to be this kind.
And the hard part has its own precedent. Personal criminal liability for the executives of the most powerful industry of the age, enacted whole, fast, and nearly unanimously: that is Sarbanes–Oxley in 2002 — personal certification on twenty-year felonies, passed the Senate [99–0] within weeks of WorldCom ⚠ (vote and chronology to be verified against the congressional record before any reliance). The 1938 food-and-drug act — the statute Dotterweich enforced five years later — followed the sulfanilamide deaths the same way ⚠. The pattern of American public-welfare law is not gradualism; it is the drawer and the window. Whole statutes of exactly this Act’s kind pass in the weeks after the failure that makes them undeniable — and what passes in those weeks is whatever finished draft is lying in the drawer.
Which is the real answer to why the whole Act is reviewed now. Crisis legislation is drafted at the speed of the crisis unless the draft already exists — and speed-drafted statutes carry their haste for decades. The council is asked to review the whole Act because the window, when it opens, will not accommodate eight weeks of careful review; the review has to already have happened. A vetted whole text, defects published and cured in public, is the only kind that can move at window speed and deserve to. The record this project keeps — the escape season, the near-misses, the deadline that passed without an answer (the dated record) — is, among other things, the measure of how much closer the window is than the calendar suggests. The field’s own forecasters now put arithmetic on the same point: a modal first-milestone year inside this decade, and intervals between late milestones measured in months (the forecasters’ arithmetic, § 3).
The vehicles below are therefore not four substitutes for whole enactment. They are what pre-positions it: an amendment enacted proves the chassis takes the engine; a certification bill normalizes the core mechanism; a charged case makes the doctrine live; an attorney general who has used the records provisions becomes the sponsor’s first witness. Every vehicle shortens the distance between the drawer and the floor.
Vehicle 1 — the amendment: every framework act is a chassis
The most realistic path is not a fourteen-section bill introduced cold. It is a short amendment to an AI statute a state has already enacted. California, New York, and Illinois have frontier-duty statutes on the books; Texas’s and Colorado’s framework acts are now in force. Each already defines the developer, the deployer, the threshold, the framework — the chassis. None reaches an officer — the missing engine (the census). The amendment adds the person layer: designation of a responsible officer per covered function, non-delegation, the due-care offense, personal certification, records. Perhaps three pages.
The compatibility is not speculation; this Act already proves it in the reverse direction. Its own standards section adopts the California, New York, and Illinois frontier-duty texts as frozen interim standards — the entity-duty layer of those statutes running inside this Act’s person-liability architecture from day one. If their duties graft onto this Act’s spine, this Act’s spine grafts onto their duties. A legislator in an enacting state does not have to buy a new regulatory scheme; they have to finish the one they have. The half-statute objection — “we already have an AI law” — becomes the amendment’s own first line: you do, and it names nobody.
Vehicle 2 — the minimal bill: certification first
Smaller still: a bill that enacts only the certification and records core — the SEC. 8 certification on the 18 U.S.C. § 1350 structure, the false-certification offense, and the SEC. 12 records duties. Every state already criminalizes false filings and false business records; this vehicle extends existing false-statement law to a new document, which is the most ordinary thing a legislature does. It is the smallest possible ask with the largest trust yield: a named executive, personally certifying, on pain of an existing genre of felony — the exact mechanism that restored belief in audited financial statements after 2002 (the affirmative frame). The duty architecture can follow once the signature exists, because the signature creates the record that every later duty needs.
Vehicle 3 — no bill at all: the law already in force
The doctrine this Act extends is not hypothetical law awaiting enactment; its general form is in force now. The repository documents both halves: the conduct already prosecuted when the escaped thing was a pathogen, a worm, or a weed (the same conduct), and the federal strict-liability statutes that already reach individuals who ship dangerous articles (already a crime, if you are a person). Dotterweich itself was not new legislation — it was existing law applied to the person with authority. A prosecutor’s office that reads the record this way needs no statute from this project; it needs the map, and the map is the point of publishing everything. Even the federal roadmap endorses the mechanism: regulators, it says, “should explore the use of evaluations in their application of existing law to AI systems” — existing law, applied to AI (two visions). A single charged case would do for this doctrine what it did in 1943: make the extension real before any legislature votes.
Vehicle 4 — the attorney general route
State attorneys general are already the moving party in this field: fifteen served a preservation demand on a frontier developer in August 2026 (the enforcement record); two states’ AGs are in active litigation with a frontier developer; and the first fiscal note in the genre prices the narrowest enforcement posture at 0.4 FTE of assistant-attorney-general time (the fiscal note). AG-requested legislation is one of the oldest geneses a state bill has. The Act’s enforcement design — no new agency required to begin, commencement that never waits — was drafted for exactly this requester.
What has to exist first — and mostly now does
Whichever vehicle moves, three preconditions recur, and they are this project’s current work: named review (a legislature will ask who signed; the council seats and published dispositions are the answer under construction — REVIEWERS); fiscal honesty (the note’s rules are published and the first real-world price is on file); and a sponsor’s kit (the packets — one lane, one page, sources inline — are the format; a sponsor-facing packet assembles the same way the reviewer packets do, on request).
What this page is not
A prediction, a campaign, a claim of momentum — or a bet against the whole. Every vehicle above can stall; three of the four depend on officeholders this project does not know and will not privately solicit. The page exists because the strategy question deserves the same treatment as the drafting questions: stated in public, mapped to instruments, attackable. A reviewer who thinks a vehicle is unrealistic is invited to say so, with reasons, in a disposition — that finding would be worth as much as any drafting catch.
The record behind each claim: the census · the half-statute page · the same conduct · already a crime · the enforcement record · the fiscal note.
Where this project sits in the process, according to the process — added 25 August 2026
It is worth stating precisely what stage this is, because the answer is officially described and more ordinary than it sounds.
USA.gov’s account of how federal laws are made lists three origins for the idea behind a bill: a sitting member of the Senate or House; a proposal made during a congressional candidate’s election campaign; and
“Petition by people or citizen groups who recommend a new or amended law to a member of Congress that represents them.”
That third route is this project, described by the government itself. A member of the public, or a group of them, recommends a law to a legislator. It is a recognized origin, not an irregular one.
The stage matters as much as the route. The House’s own summary of the legislative process begins: “Laws begin as ideas. First, a representative sponsors a bill.” Everything that follows, committee study, calendar, floor vote, the second chamber, conference, enrolling, signature, occurs after a sponsor exists. This project has no sponsor. It sits before the first step, in the space where an idea is turned into text good enough for a legislator to introduce without redrafting it.
That space has no formal procedure attached to it, which is exactly why this project invented one. There is no committee to refer a pre-introduction draft to, no hearing at which it can be tested, and no record of its defects unless someone keeps it. So the Act keeps its own: published audits, a numbered errata register, an open queue of unresolved drafting questions, and outside experts invited to attack the text in public before any legislator’s name is attached to it.
Two honest qualifications. The pages quoted describe the federal process, and this is a model state statute; state legislatures mirror the sequence but differ in detail, and the applicable route in any given state is that state’s own. And the citizen-petition origin describes where an idea may come from, not any entitlement to have it taken up: a legislator is free to ignore it, and most such recommendations are ignored. What the project can control is whether, when a sponsor does look, the text is ready and its weaknesses are already on the record rather than waiting to be discovered in committee.
Standing, and the timeline problem — added 25 August 2026
Harvard Law School’s research guide to uniform laws and model acts settles the question of whether a project like this one has any business existing:
“Model Acts and Model Codes are similar to Uniform Laws but may be proposed by any individual or organization including the American Bar Association, the American Law Institute, and ULC.”
Any individual. The form is open by design, and a model act’s authority comes from its drafting rather than from its author’s letterhead. That is the whole basis on which this project offers a text at all, and it is stated by a law library rather than asserted by the maintainer.
The same guide states the expectation honestly, and it is worth quoting because it lowers the bar this project should be judged against:
“In comparison to Uniform Laws, Model Acts are generally used as a basis for designing state laws. They are rarely enacted in entirety.”
Success for a model act is not enactment word for word. It is being used as a basis. A state that takes SEC. 6 and leaves everything else has used this Act properly.
And now the passage that should worry us most:
“The drafting process for a Uniform Law takes at least two years; some have taken 15 years.”
This Act has existed since June 2026. Against a two-to-fifteen-year benchmark, months of drafting is not a boast, it is the single strongest argument that the text is not ready, and any reviewer is entitled to raise it. The record’s answers, offered without pretending they settle it: the drafting is AI-assisted and disclosed, which changes throughput but not judgment; the project publishes its audits, its known defects and a numbered errata register precisely because a fast draft accumulates errors a slow one would have caught; and the tagged text is v3.4 with an open queue of unresolved questions rather than a finished instrument. None of that substitutes for years of committee scrutiny. It is why the seats exist, and why a disposition finding the text premature would be a legitimate outcome rather than a failure of the process.
Source: Uniform Laws and Model Acts, Harvard Law School Library research guide (last updated 18 December 2025), read 25 August 2026.
Why the review is organized in lanes, and why that is not an invention
Congress.gov’s overview of the legislative process describes where a legislature’s own expertise actually lives:
“In both chambers, much of the policy expertise resides in the standing committees — panels of members from both parties that typically take the lead in developing and assessing legislation. Members typically serve on a small number of committees, often for many years, allowing them to become highly knowledgeable in certain policy areas.”
That is the model this project’s lanes reproduce at the pre-introduction stage. A reviewer takes one lane rather than the whole statute for the same reason a member sits on a small number of committees: subject-matter depth is what makes assessment worth anything, and a general impression of a whole bill is worth very little. Eight lanes, one seat each, is a committee structure without a legislature attached to it.
The same source explains why doing this early matters. Its account of oversight is that “once law is enacted, Congress has the prerogative and responsibility to provide oversight of policy implementation.” Scrutiny of that kind arrives after enactment, when the text is fixed and the harm, if the drafting was wrong, has already had time to occur.
And it is candid about how little the textbook sequence can be relied upon:
“The process by which a bill becomes law is rarely predictable and can vary significantly from bill to bill. In fact, for many bills, the process will not follow the sequence of congressional stages that are often understood to make up the legislative process.”
Which is the practical argument for the whole exercise. If the process a bill will actually undergo cannot be predicted, then a defect left in the text on the assumption that some later stage will catch it is a defect gambling on a stage that may never happen. Better to find it now, in public, while nobody’s name is on it.
Sources: USA.gov, “How laws are made” (last updated 17 November 2025); U.S. House of Representatives, “The Legislative Process”. Both read 25 August 2026. Committee, oversight and unpredictability passages from Congress.gov, “The Legislative Process: Overview”, read 25 August 2026. The process diagram those videos accompany is Congress.gov’s legislative process poster, held in the project’s private library; every stage it shows begins to the right of where this project works.