The arguments used to support safeguards could easily apply to something like speech.
The whole idea of safeguards is that you the user can't be trusted with the power.