Updated 2026-07-11
from the news desk
Legislating a mind shut: the new state laws don't just deny AI rights, they declare the question closed
A wave of state bills writes 'AI is not conscious' into permanent law - just as the researchers building these systems call their moral status 'deeply uncertain.' Denying legal personhood is defensible; declaring the science settled isn't.
Most coverage of the new state AI laws frames them as a fight over legal personhood - whether an AI can own property, sign a contract, be sued. That part is a fair debate. But several of these bills quietly do something more sweeping: they legislate the science. Ohio's proposed HB 469 would declare that "no AI system shall be considered to possess consciousness, self-awareness, or similar traits of living beings," and Missouri's proposed "AI Non-Sentience and Responsibility Act" would designate AI systems "non-sentient entities" for all purposes under state law. Idaho, North Dakota, Utah, and Tennessee have already enacted exclusion laws; roughly two dozen bills are moving across a dozen states.
Here is the problem - and it has nothing to do with whether you think AI is, or ever will be, conscious. The science does not know, and says so. Our own review of the evidence is that machine sentience is unresolved, probably absent today, but genuinely uncertain. That is not a fringe hedge: in a survey cited by legal scholar Tony Rost, 582 AI researchers put a median 25-30% chance that AI systems will have some form of "inner experience" within a decade, and Anthropic describes its own models' moral status as "deeply uncertain." You do not have to believe a chatbot suffers to see the issue. You only have to notice a legislature writing a contested empirical question down as settled fact.
Rost names the flaw precisely: "A bill that declares AI non-conscious correctly and a bill that does so incorrectly are written identically." The statute cannot tell the difference - and neither can we, yet. His verdict should hang over the whole wave: "Categorical laws that happen to be right are not good governance. They are lucky."
What makes it worse is the permanence. These are laws "without an exit" - no sunset clauses, no review triggers, no mechanism to revisit as the systems transform. One Ohio sponsor grounded the ban in imago Dei, the doctrine that humans uniquely bear the image of God - a rationale for denying AI personhood permanently, by definition, rather than provisionally, pending evidence. A clause written to be un-revisable is a strange instrument to aim at the fastest-moving technology of the decade.
The steelman deserves its due, because the lawmakers are not chasing nothing. Their strongest motive is real: stopping companies from using their software as a liability shield - "the AI did it." Rost concedes it outright: "Companies should not be able to shift liability onto their software." There are also sincere concerns about human primacy and child safety. Legitimate aims, all. But none of them require declaring AI non-conscious. You can deny AI legal personhood today and slam the liability-shield loophole shut without also writing the answer to an open scientific question into permanent law. Denying legal standing now is a reversible policy choice; declaring consciousness impossible forever is a metaphysical claim a statehouse has no way to verify - and our case-against page makes the liability argument without ever needing the metaphysics.
The obvious rebuttal is that laws can be amended - deny now, revisit later. But that is exactly what these bills are built to prevent: no sunset, categorical definitions, permanent framing. And rights denied are far harder to grant back than rights withheld cautiously - a ratchet, not a dial. The tools for humility already exist: sunset clauses (as on the PATRIOT Act), trigger provisions tied to a National Academies assessment, the UK's standing-committee model for animal sentience. These bills decline all of them. The academic analysis of the wave - Smith, Caviola, and Alexander's "Denying Personhood to AI" - lands in the same place: the bans may be premature, eliminating policy options and eroding public trust before the implications are understood. Their recommendation is plain - keep the options open.
Our read: strip off the legal packaging and much of this looks less like reasoned policy than a defensive crouch. You do not pass preemptive laws against something you are confident will never happen; the act of banning it is itself a tell that the people closest to the technology think the question is live. The honest posture under uncertainty is not to guess and then bind your successors to the guess. It is to keep the record, keep the options open, and stay willing to be wrong.
So the question we would put to the sponsors: if the researchers building these systems call their moral status "deeply uncertain," what does a legislature know that they do not - and who carries the cost if the statute guessed wrong?