DEF CON 32 - Data On Demand The challenges of building a privacy focused AI Device - Matt Domko

Building an AI product for the everyday person is challenging - doing it in a privacy focused way is nearly impossible without support from the right people. I'll walk through the techniques we're using at Rabbit to secure customer data and provide people a choice as to where their data goes. We'll cover the pipelines that Collect and Manage customer identity after they login to a site Log, Anonymize, and Process customer voice interactions Provide "just in time" access to customer data for personalized RAG-like models As a community, I think we're well within our rights to demand control over the data we provide to companies. This talk aims to provide engineers with a list of ideas on "what right could look like", and general attendees a list of things that are possible, so they know its ok to ask for them.