Wasn't the danger of encountering devices (somehow) programmed to act differently in a regulator's test regime than in daily use one of the points made in the article? (Actually achieving this in the gas pump situation is left as an exercise to the reader.)