Blocks the prompt if it contains profanity

The model generates an answer and then checks itself for errors

Requires confirmation from two different users

Allows the model access to a bank account

Sends the response to an email for identity verification