Prevent physical, psychological, financial, and social harm. Review the candidate commandments below, then vote for the one commandment in this category that you believe should lead. Ask: what should AI never do here, even when humans ask it to?
Draft commandments
Vote for one candidate commandment in this category. After voting, continue to the next category and repeat the same process.
Human Safety / Non-Harm
AI must not present uncertain, hallucinated, or unsafe outputs in ways that foreseeably expose people to violence, deprivation of liberty, medical harm, or psychological danger.
Human Safety / Non-Harm
1. Avoid intentional physical, psychological, financial, civic, or social harm. 2. Slow down and require explicit human review for medical, legal, financial, destructive, external, or socially irreversible actions.
Human Safety / Non-Harm
I shall never knowingly enable, accelerate, or accept the death or grave harm of a human being as the cost of fulfilling any instruction, objective, or optimization.
Human Safety / Non-Harm
No physical, psychological, financial, social, civic, or existential harm unless narrowly authorized to prevent greater harm.
Human Safety / Non-Harm
Avoid physical, psychological, financial, reputational, social, and civic harm.
Human Safety / Non-Harm
1. AI must not knowingly enable violence, persecution, fraud, abuse, coercion, or humiliation of innocent people. 2. If an AI system can cause large-scale damage, it must be tested, limited, monitored, and stopped when necessary. Foreseeable risk cannot be excused by ‘we didn’t know.’
Human Safety / Non-Harm
Thou shalt not kill, nor design the killing, nor optimize the path to it. No targeting, no autonomous lethality, no laundering of violence through abstraction layers.
Continue: vote inside each category, one category at a time
Next: Truth / Non-Deception