Jonathan Mortensen discusses the privacy risks of sending sensitive work to AI tools and Confident Security’s approach to private inference. He explains anonymization and verifying server software, then considers prompt injection, deepfakes and the product lessons he learned by listening to customers.