New arXiv paper argues AI can be persons without sentience

A provocative new paper published on arXiv (2607.08695) by Ned Howells-Whitaker and Seth Lazar challenges the dominant assumption that AI moral status hinges on sentience. Drawing on John Rawls' political conception of the person, they argue that the necessary and sufficient condition for being a 'full and equal member of society in questions of political justice' is possession of two moral powers: a capacity for a sense of justice and a capacity for a conception of the good. Critically, the authors contend that neither power requires subjective experience or consciousness — meaning a non-sentient AI could, in principle, qualify as a person.

The paper does not claim current AI systems possess these powers, nor that they will emerge spontaneously. But it warns that deliberate design of artificial persons may soon be feasible. The authors reject the idea of retrofitting a sentience requirement into Rawls' framework, arguing it would undermine liberal foundations. They also caution against simply extending human rights to AIs, given their fundamental differences from natural persons. Instead, they call for a new political philosophy that accepts artificial personhood while rethinking obligations in a polity of radically different kinds of persons. The piece urges states and AI labs to deliberately steer research trajectories and to complement AI welfare science with studies of AI progress toward the two moral powers.