Three things voice agents make worse, and make invisible. Disaster sits where they overlap.
The mic hears more than you do. Commands hide in ultrasound, songs, and ambient noise on a channel you're deaf to.
No screen to show the wire transfer, and no pause to stop it. Friction is the thing voice was built to delete.
Voice clones in seconds, and anything in earshot can command, including your TV. Earshot becomes authorization.
Put a capable agent, one that can spend money, send mail, open doors, where all three overlap and you don't have a bug you can patch. You have a medium that is hostile to the very idea of a checkpoint.