If your AI alignment strategy is so fickle that it breaks if people simply discuss potential problems with the strategy then you didn't really have an alignment strategy to begin with.