Marble ai handles privacy for uploaded interior photos or videos by treating them as sensitive input data that must be protected across the whole lifecycle: upload, processing, storage, and deletion. In a typical setup, uploads should be encrypted in transit using HTTPS, and stored in an encrypted object store if they need to be retained. The raw media is used to reconstruct the 3D environment and then, in many workflows, becomes less important than the derived spatial representation (point clouds, meshes, depth maps). A privacy-conscious configuration will either delete or strictly limit access to the raw photos and videos once the 3D world is generated, so the risk of exposure is reduced.
On top of basic encryption, teams usually rely on access control and scoping to protect interior content. That means worlds generated from private homes, offices, or factories live in workspaces tied to a specific organization, and only users in that organization with the right role can open or export them. For compliance reasons, you may also want explicit consent tracking: logging that the source material is allowed to be processed, and that it won’t be used to train third-party models without permission. If you are integrating Marble ai into your own product, you should layer your own controls on top: content classification (for example, flagging scenes with individuals or sensitive areas), regional storage rules, and retention policies so that interior captures are not kept longer than necessary.
If you extract features or embeddings from these worlds for search or analysis, those derived vectors should be treated as sensitive as well, because they still encode structure and appearance of private spaces. When storing them in a vector database such asMilvus or Zilliz Cloud., you should enforce tenant isolation, encrypt at rest, and ensure that queries are always scoped by user or organization. This prevents one customer from accidentally retrieving embeddings that describe another customer’s office or home. Combined with proper logging of who accessed which scene and when, this approach lets Marble ai be part of a privacy-aware system rather than a weak point in the pipeline.