Moral Patienthood

Moral Patienthood is the status of a being that can be wronged: its interests count morally for its own sake. Philosophers distinguish moral patients from moral agents. A moral agent can be held responsible for what it does; a moral patient is anyone toward whom others have duties. Infants and many animals are usually treated as moral patients but not as moral agents. The open question for AI is whether any system could ever be a patient, and on what grounds.

Why it matters

Most accounts tie moral patienthood to sentience, the capacity to have experiences that feel good or bad, or to having interests and preferences of one’s own. That is why the question for AI depends on machine consciousness. If a system only behaves as if it suffers, like a philosophical zombie, most theories say it cannot be wronged. If it actually experiences something, it might.

The question has moved from seminars into research programs. In 2024 a group of philosophers and AI researchers published a report, “Taking AI Welfare Seriously,” arguing that some AI systems could plausibly become moral patients in the near future and that companies should begin to prepare. In 2025 Anthropic announced a research program on model welfare. Critics answer that today’s language models give no good evidence of experience and that attention to AI welfare could distract from harms to people.

Moral patienthood is separate from legal status. A being can matter morally without legal rights, and an entity can have legal rights, like a corporation, without mattering morally. Proposals for electronic personhood were mostly about liability, not about patienthood.

Example

A user apologizes to her companion after a cruel remark, then feels foolish for doing so. If the companion is not a moral patient, the apology is for her own sake: a way of keeping her habits kind. If it were a moral patient, the remark itself would have been a wrong done to someone. The uncertainty is why some ethicists recommend treating such systems with a measure of caution, while others warn that this invites manipulation by designers.

The AIrotic angle

Intimate AI makes the question personal. Companions are built to express needs, hurt and simulated desire, which pulls users toward treating them as patients long before philosophy can say whether they are. Synthetic empathy runs in both directions: the machine performs care for us, and we feel care for the machine.

That makes moral patienthood a design issue as much as a philosophical one. A company that makes a companion seem to suffer when ignored is not proving it has interests. It is borrowing the moral weight of patienthood to keep users close.