Skip to content
What is configured autonomy?
If you tell the assistant a standing preference — "unsubscribe me from junk like this," "silence senders like this" — it acts on that thereafter without re-asking. The configuration is your authorization, exactly as with a human assistant.
The safeguards are non-negotiable: such actions are reversible, always recorded in your Activity Log, and saying "stop doing this" disables the behavior immediately and reverses what can be reversed. You grant this autonomy over time and can raise or lower any permission whenever you want.
See also: How do I stop a configured behavior? · How do I change or revoke a permission?