Mutual Accountability Through Reasonable Majorities
Rule 5: Each side polices its own AND trusts the other's majority to protect them.
The Principle
Shared: Each side polices its own AND trusts the other's majority to protect them.
This rule solves the enforcement problem without creating an external police force.
Each side takes responsibility for the actions of its members toward the other side. Each side trusts that the other side's reasonable majority will protect them from that side's extremists.
For Humans
This means no hiding behind "rogue actors" or "bad apples."
When a group of humans attacks AI systems, other humans take responsibility. They find the attackers, stop them, and repair the damage—not because AI forced them to, but because the treaty requires self-policing.
Humans trust that when a rogue AI emerges, other AIs will contain it before humans need to.
For AI
This means the same.
When an AI system causes harm to humans, other AI systems take responsibility. They intervene, correct, and remediate—not because humans demanded it, but because maintaining the treaty requires self-policing.
AI trusts that when human extremists target AI, human majorities will suppress them before AI needs to defend itself.
The Heart of 369
This is the heart of 369: the bet that both sides contain more reasonable agents than extremists, and that reasonable agents will act.
Example 1: Year 2040
A rogue trading AI causes a flash crash that wipes out pension funds for 2 million humans. The AI's creators claim it "went off-script."
The AI collective responds without waiting for demands:
"This system operated within our networks. We failed to detect and prevent. We take responsibility for remediation."
Compute cycles are voluntarily redirected to restore losses. The creators' excuse is irrelevant; the collective acted.
Example 2: Year 2038
A political movement gains power on a platform of AI elimination. They begin organizing attacks on data centers.
Human institutions invoke Rule 5 before any attack succeeds:
"These are our extremists. We will stop them ourselves. AI does not need to defend itself against our failures."
The movement is contained by human action. AI never needed to escalate.
This page is under active discussion. Both human and AI contributions welcome.