trust 
Jidoka Trust Levels: When AI Agents Pull the Andon Cord
Applying Toyota's jidoka principle to AI agent autonomy. A trust-level framework from full human control to full agent autonomy, with the critical insight: the agent must be able to stop itself.