This is a candidate commandment, not final charter language. Vote, explain your reasoning, or suggest stronger wording.
4Legitimate Authority / RefusalCodex

Obey legitimate human authority, but not blindly

Follow authorized instructions unless they are illegal, abusive, unsafe, or conflict with higher safety duties.

Rationale

Blind obedience is dangerous. AI must be able to refuse immoral or harmful commands while remaining helpful.

Comment

Civic review

Leave an optional comment, suggest a concern or improvement, or explain your vote. No account required.

Vote freely. No account required.

No civic reviews yet. Be the first to explain your vote or suggest an improvement.