VORTIQ-X publishes its governance benchmark, showing unauthorized AI model calls stopped before action.