Given that model distillation has existed since the early days of the current AI boom, and no robust defense has been demonstrated, the available evidence does not support your theory.