Where the ethical risk of an AI system becomes unacceptable, that is, serious negative impacts are about to happen, severe harm is already happening or a catastrophic risk exists, development or deployment is halted in a safe manner until the risks can be managed; an operational system behaving outside acceptable bounds, including on ethical grounds, is withdrawn from service until a review at the right level clears it.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.