Updated 2026-07-10
from the news desk
Statehouses are voting AI's mind closed - just as the labs learn to measure it
Tennessee wrote AI out of the legal definition of a 'person,' and more states are following - even as welfare researchers treat the underlying question as probabilistic, multidimensional, and maybe unanswerable.
Tennessee has written artificial intelligence out of the definition of a "person." Under SB 837, now signed into law, a "person" in the state code no longer includes "AI, a computer algorithm, a software program, computer hardware, or any type of machine." It is not an outlier: Idaho, Utah, and North Dakota have already passed measures denying AI legal personhood, with Oklahoma and Ohio close behind.
The stated reason is blunt, and on its own terms serious. Oklahoma's Rep. Cody Maynard says AI "is a man-made tool, and it should not have any more rights than a hammer," and the animating fear is accountability — that without a hard line, companies could "shift off any accidents onto the AI." That is not a frivolous worry: legal personhood for AI really could become a liability shield, which is one of the strongest points in the case against rights.
What makes the timing worth a second look is what is happening in the labs at the same moment. The researchers whose actual job is this question are moving in the opposite epistemic direction. Robert Long, of the AI-welfare research group Eleos, lays out how to study AI welfare empirically — treating it as probabilistic and multidimensional, weighing evidence about "welfare grounds" and "welfare interests" under open uncertainty, and acting on precaution rather than waiting for a verdict. Cambridge philosopher Tom McClelland argues the verdict may never arrive at all: "The logical position is agnosticism. We cannot, and may never, know."
Our read: the law is settling by statute the exact question the science says cannot be settled. In principle there is no contradiction — legal personhood and moral patienthood are different things, and a legislature can deny the first while the second stays wide open. The move worth watching is when a bill goes further and declares AI non-conscious as a matter of code, not merely non-person as a matter of law. A statehouse can define a legal term however it likes; it cannot legislate an empirical fact into being — and the people closest to the systems are telling us that fact isn't in yet.
So the question we would put to both sides: if science genuinely cannot tell us whether a system can suffer, should the law be allowed to declare that it cannot — and who is accountable if the statute turns out to be wrong?
Sources
- Troutman - Proposed State AI Law Update (Tennessee SB 837 signed into law) — 2026-04-27
- Straight Arrow News - States rush to deny AI personhood — 2026-02-02
- University of Cambridge - We may never be able to tell if AI becomes conscious, argues philosopher (Tom McClelland) — 2026
- Experience Machines / Eleos (Robert Long) - promising research directions in AI welfare — 2026