Davidmanheim comments on Don’t design agents which exploit adversarial inputs