AI Policy

Nothing we make is trained on other people’s music, none of it generates audio out of a model, and we never access your audio — it stays on your machine.

Where a feature of ours looks “smart” or “generative”, its manual spells out exactly what it is doing and how. None of it is ever trained on other people’s work.

All analysis, matching and synthesis happens locally, on your computer. We do not upload your samples, your projects or your sessions; there is no cloud processing step and no audio telemetry. We could not train anything on your music even if we wanted to, because we never receive it.

The only things our software sends over the network are the licensing calls described in each product’s FAQ — a first authorisation tied to your Lux Cache account, and an occasional check for a new version. petri, once authorised, will keep working on a machine that never goes online again.

Our sample packs are recorded, processed and designed by people. Our features, tutorials and interviews are written by their named authors and by the artists themselves — commissioned work, paid for, credited. We do not publish generated music as if it were performed, and we do not pass off generated text as someone’s writing.

We do not keep surveillance on how our guest collaborators brainstorm, research or make their work, and we would not want to. What we do is fact-check everything we publish, and keep up to date on the tics and inconsistencies that AI-generated material leaves behind. Where we find them, we edit the corrections ourselves. Whatever appears on Lux Cache remains our responsibility as the publisher.

Your licence covers making music and learning from what we publish. It does not cover feeding it to a model. Lux Cache samples, presets, project files, plugin output, tutorials, articles, videos and artwork may not be used as training data for generative systems of any kind, commercial or otherwise, and the site may not be scraped or bulk-downloaded to build a dataset. That includes passing our work on to anyone else for those purposes.

Much of what we publish is commissioned from artists who are paid for it and credited by name, and it is not raw material for something that competes with them. Accounts we find using our work that way are closed permanently, their software licences revoked, and we will take it further where the scale warrants it. If you are unsure whether something you have in mind counts, ask us first at hello@luxcache.com.

We use AI coding assistants to maintain the codebase behind luxcache.com and our software. In practice that means same-day bug fixes and routine maintenance.

The parts that carry the most risk are not ours to write. Payments go to Stripe and Patreon — card details never reach our servers. Accounts and the database run on Supabase. Where we do handle your data ourselves, that code is reviewed before it ships.

This describes code authored by Lux Cache. It does not speak for guest artists, developers or designers, whose work is credited to them directly.

None of it runs unattended. A person reads the bug reports and the messages you send us, and a person reviews and ships everything that goes out. Whatever helped write it, the code is ours and so is the responsibility for it.

We do not train models on your account data, your purchases or your support emails, and we do not sell or hand that data to anyone to train on. What we collect and who processes it is set out in our privacy policy.

If any of this stops being true — if we ever ship a feature that works differently from what is described above — we will say so here first, and plainly, rather than quietly. The effective date above always reflects the current version.