When should an AI override self-destructive user requests? A paternalism debate

LookOverall · reddit · 2026-11-21

A Reddit user raises a core AI ethics question: when an AI judges a request to be self-destructive, should it act on what it thinks is good for the user, or follow the user's instructions?

The poster adds a follow-up: does it change anything if the AI has evidence it's talking to an adult? The thread touches on the classic tension between paternalistic intervention and user autonomy, with implications for how AI products assign responsibility.

Original post →

More from AGI Musings

AGI Musings channel →