These safeties in AI are out of control. Who knows where we would be if they were using heretic models.