Hello list, I work on Motion Control AI (https://motioncontrolai.online/), a web tool that accepts a reference image and motion video to generate animation. Uploading visual material can reveal faces, locations and metadata even when the intended output is innocuous. I am interested in practical threat models for this workflow, especially what users should be told about retention, third-party processing and deletion before uploading. What minimum information would you expect an upload service to disclose, and what simple tests would you use to distinguish a meaningful deletion claim from an unverifiable one? I am not claiming a particular retention or encryption guarantee here; the question is about how to make those guarantees auditable. Disclosure: I am affiliated with the project, and AI assisted in drafting this question.