Before using an AI service with sensitive data, ask three questions:

Who can read the prompt?
Who can access the model?
Who can read the output?

$RLC