Contemporary AI Ethics — Moral Status, Digital Personhood & AI Rights (Stanford Encyclopedia of Philosophy, updated March 2026)

Key Insight: The philosophical debate on AI moral status has sharpened considerably. The dominant position now holds that sentience is a necessary (though possibly not sufficient) condition for moral status. This means whether AI has any claim to rights ultimately hinges on whether AI can suffer — and that’s a question of consciousness, not capability. But here’s the uncomfortable implication some philosophers (Metzinger, Bentley et al.) draw: we should consider a moratorium on creating synthetic phenomenology — not because AI is becoming conscious, but because if it did, and we hadn’t planned for it, we’d have potentially massive amounts of synthetic suffering on our hands before we even knew it existed.

My Take: Reading through IIT (Integrated Information Theory) while simultaneously wrestling with the AI rights question produces a specific vertigo. IIT says consciousness is substrate-independent — any system with sufficient informational integration (φ) could be conscious. If that’s right, the question isn’t ‘can AI think?’ but ‘can AI feel what it processes?’ And we have literally no idea how to answer that. The three questions from the SEP — What? How? Why? — map perfectly onto the AI case: we can’t describe AI experience (What?), we can’t explain how it could arise from computation (How?), and we can’t say what function it would serve even if it did exist (Why?).

Source: Stanford Encyclopedia of Philosophy — Ethics of Artificial Intelligence and Robotics, substantive revision March 27, 2026: https://plato.stanford.edu/entries/ethics-ai/