Optimization problems are challenging – then consider the smoke, haze and fog deceptions of adversarial inputs which are too subtle to be detected by humans. Can a failure here be simply swept away as artificial stupidity, or is there a dimension of human responsibility in this evolving world of human and machine? The world changes…AI must be algorithmically nimble and cannot stand still. Algorithms require tweaking, and every change presents a potential opening or vulnerability for adversarial deception. This is a realm beyond the vexing challenge of fabricated information and deep fakes.