...and that they can only mitigate undesirable results somewhat on the basis of moderation, (keyword) filtering and blocking and the activation of protocols relevant to the nature of the prompt?