Private files should not become AI training data

AI is already part of modern work. It can help people draft, search, summarize, translate, review, and organize. Sandra is not built on fear of AI, and we are not interested in pretending useful tools are the enemy.

The line is consent. A client file sent for delivery is not an invitation to train a model. A legal bundle, unreleased campaign, financial export, architectural drawing, music master, or private archive should not quietly become training data because it passed through a file service.

When someone sends a file through Sandra, they are usually trying to complete a specific job: deliver the work, collect missing documents, share a private link, or keep an important file in Vault. The file has a purpose. It is not there to be mined for ads, profiled, indexed, or fed into a model.

That is why Sandra keeps the promise plain: customer files are private content. Sandra does not use uploaded files, transfers, file requests, or Vault files to train AI models.

Law firms and legal teams handle files with a different level of responsibility: contracts, exhibits, discovery material, signed documents, negotiation drafts, identity documents, and privileged communications. The risk is not only a bad user experience. It is confidentiality, professional duty, and trust.

A private transfer workflow helps reduce accidental exposure. Expiring links, password protection, controlled downloads, and a private Vault are more appropriate than loose attachments or broad shared folders when the document should reach a specific person and no further.

Photographers, filmmakers, designers, musicians, architects, and agencies often send work before it is public. Those files can include originals, rejected concepts, client strategy, unreleased material, source files, and private feedback.

Sandra treats that work as private too. The fact that a file is visual, artistic, or commercial does not make it free training material. The owner decides what happens to it.

Sandra focuses on file delivery: send files, request files, and keep important files in Vault. Files are protected in transit and stored encrypted at rest. For transfers that need an extra layer, Sandra can use browser-side end-to-end encryption.

The product stays useful without crossing the line. No ad profiling. No private content indexing. No AI training on customer files. Just a clear place to move important files with less friction.

That position is simple enough to explain to a client: we can use modern software without handing your documents to a training pipeline.