@kensanata @brainblasted the thing that worries me is when someone works out how to get the AIs to launch a huge pretexting / social engineering campaign. Human voices make everything seem more realistic and "trustworthy" (for now), especially for less savvy people. It could even go further and:
1. Automatically call for a "survey"
2. One week later, call using a different voice. Say there is a problem with the bank account
3. Automatically call the bank and empty the account