When an autonomous AI agent's actions diverge from intended operational objectives due to adversarial influence.