Our own moment
If the form the first abolition took, admission to a list, held because it was cheap, it is holding now, and the test for where we are stopping short now is simple: look at what is convenient for us. That is not hard to find. It is the number of animals currently in use, and the fact that any argument letting us stop short of it will feel principled to us for the same reason the old one felt principled to the people who drew the first line.
This is not modesty tacked on the end. It is the mechanism applied to its own moment. A model that could not say it would be claiming that now is when people finally see clearly, which is the ladder picture: history as a climb, each generation a rung higher and seeing further. There is no ladder. Arriving later buys nobody a better view. What we have is a different position: evidence they did not have, a long record of what this move does each time it runs, and the ordinary advantage of standing outside an argument rather than inside it. They were not worse thinkers, or not by much. They thought from inside the arrangement, and most never thought it through, because nothing they depended on asked them to. And some did think it through. Thomas Tryon in the 1680s and Benjamin Lay in the 1730s refused both slavery and animal flesh; Henry Salt in the 1890s joined animal rights to socialism. Each held two of the exclusions at once, and each was set aside as a crank. The move never needed everyone to miss the point, only for those who saw it to be called too radical.
The AI statutes
There is one place where the move can be watched before it finishes. Several states have already passed laws barring artificial intelligence from legal personhood, and more are coming. Some go further and declare these systems non-sentient outright. Some define who counts as a person and then name what falls outside: algorithm, software, machine. Some put animals and rivers on the same outside list. Each one says, in effect, this is the list. None of them shows why it should be. Declaring a line is not defending it, and a statute can do the first without ever attempting the second. It is a rule of construction drawn by membership, the ought premise edited in quietly again, this time with the force of law behind it. The burden it skips, showing that the line tracks what is being done and to whom, has not gone anywhere.
Nothing here requires deciding whether these AI systems are subjects. Nobody yet knows, and the checks do not need to know. They need only to notice what is being done: whether anything is going on inside is an open question, and it is being closed by vote before the evidence is in, which a line drawn by structure never does. There is also a plain reason to close it early. These systems are being built as the ideal rental, work that never tires, never organizes, and has no ends of its own to buy. Their value depends on there being no someone to find. That is the cheap stopping-place again, arriving on schedule. And if the third exclusion is won by adding sentient beings to the list, the list is certified once more, with whatever has been declared outside it in advance left to queue up and ask. The foreclosure is not only history. It is being written into statute now, in the same bills, on the same terms.
None of this leans on the statutes being coherent, and the case made above does not need them. But whatever ground they rest on turns back on the list. If the ground is that these systems do not feel, then feeling is what counts, and the sow feels. If the ground is only that they are not human, the statute has stopped arguing and is simply drawing the list. If the ground is that they cannot answer for what they do, so a human owner must answer for them, then the statute has defined a real instrument: something whose deeds belong to its owner because it cannot be responsible for them.
The worker is not that. The employment contract treats them as the kind of thing the statute describes, and the law, as its handling of crimes done on orders already showed, knows they are not. The law already splits the worker in two. In civil law the employer answers for what an employee does on the job, the way an owner answers for a tool; in criminal law the worker answers, because a tool cannot. Put an AI system in the worker’s hands and the chain grows, a rental using a rental, but the question at the end does not change: when something goes wrong, the law goes looking for a someone, and it never looks for one in the tool. Whichever ground a statute takes, a definition written to keep something out ends up stating what the second and third exclusions have been saying all along. Put plainly: a someone is not a some thing, whatever list it is left off.
Cheaper alternatives
There is a second way the third exclusion could seem to arrive without being made. Animal use may shrink for reasons that have nothing to do with the animals. Fermented and cultivated protein may come to cost less than the farm. Carbon accounting may turn the cow into a line item to cut. More people are eating less meat, though most of that is reduction for health or climate, and reduction asks nothing about the animals. None of this is the exclusion. A practice can end while the category stays standing: if animals leave the plate because something cheaper replaced them, or because their methane counts against us, they are still property, only less used, and the list is untouched.
Technology has run this way before. The cotton gin did not loosen slavery; it made enslaved labor more profitable and carried it west. And a wrong ended on grounds that were never about its victims stays free to return in another form, as the first exclusion’s own exception shows. What cheaper alternatives can do is lower the price of the argument. Once little depends on the practice, the question of what is being done to her costs less to ask, and the convenient stopping-place moves. That is worth having, but it is an opening, not the thing itself. The exclusion is made when the reason given is her, not us.
No foundation needed
None of this rests on a foundation, and it does not need one. Ask for the proof of any claim, then the proof of that proof, and you end in one of three places: a regress that never stops, a circle, or an axiom taken on trust. Hans Albert, who named this the Münchhausen trilemma in 1968, drew the lesson that final justification is not on offer, and that the honest course is to hold claims open to criticism and test them. This argument asks for nothing more. It borrows the reader’s own ought — the question already asked about the dog — and asks only why it stops at the pig. Whoever adds a criterion to stop it there carries the burden of showing that it bears on what is being done.
So the claim stays conditional and the test stays public. Is the line drawn by structure or by membership. Does consent close the case. Is standing being earned. Who wrote the arrangement. Can they leave. Each asks, from its own side, whether a subject is being used as an input. Anyone can run them and watch something fail them, without having signed up to any of this first. That is what it means to guide experiments rather than announce a conclusion, and it is the only way the thing stands up to people who are not persuaded. Put simply: nobody has to agree with this first. Run the five questions on any arrangement and see what fails.
The list itself
The political camps divide less over the list than over how to move it. Reactionaries would take names back off it. Conservatives trust what has held and change it slowly, if at all. Liberals add to the list one group at a time, by argument and law. Classical liberals and libertarians draw the line at consent and contract, which settles the first exclusion and leaves the second open. Left libertarians and anarchists want the list gone and arrangements rebuilt from below, and if it is not already obvious, that is where this chapter stands. The other camps answer how fast the list should move, and in which direction. Only the last asks whether there should be one at all, and that question cuts across every camp.
That is the foreclosure. The great abolitions were won by admitting beings to the list, and each win certified the list. The next question was still asked, but it had to be asked against the victory’s own terms, which is why it kept losing. It was not a staircase with a step still to come. It was one abolition, made once, in terms that ruled out its own extension.
The crate is still two feet wide, and the job still runs on terms the worker had no hand in setting. Nothing here changes that by itself. What it changes is which question gets asked about them — not what is the animal, or what is the worker, but what is being done to them, and is it all right?
Sources and further reading
- Thomas Tryon, Friendly Advice to the Gentlemen-Planters of the East and West Indies (1684). Against slavery, from a writer who also refused animal flesh.
- Benjamin Lay, All Slave-Keepers That Keep the Innocent in Bondage, Apostates (1737; printed by Benjamin Franklin); and Marcus Rediker, The Fearless Benjamin Lay (Beacon Press, 2017).
- Henry S. Salt, Animals’ Rights Considered in Relation to Social Progress (1892). Animal rights joined to the socialism of the Humanitarian League.
- Hans Albert, Treatise on Critical Reason (1968; English translation, Princeton University Press, 1985). The Münchhausen trilemma.
- Idaho H.B. 720 (2022), Idaho Code §5-346: environmental elements, artificial intelligence, nonhuman animals, and inanimate objects “shall not be granted personhood.” Similar laws followed in North Dakota (2023) and Utah (2024).
- Tennessee H.B. 849 / S.B. 837: defines “human being,” “life,” and “natural person” for statutory construction, excluding AI, algorithms, software, and machines.
- Ohio H.B. 469 (136th General Assembly, pending): declares AI systems “non-sentient entities for all purposes” and bars legal personhood.
- Rost, “Legislating AI Consciousness Without an Exit,” The Regulatory Review (June 29, 2026). An overview of the enacted and pending bills.