In AI terms, you could think of Frodo as an adversarial example, specifically chosen to exploit a known bias in the watcher’s model.