We commit to a public Data Trust document that codifies how the corpus is built, used, and constrained. It is not a privacy policy. It is a governance instrument — reviewable by federation legal, auditable by federation safeguarding officers, and binding on us regardless of who owns SchoolFish in the future.
The document is in development, with publication targeted for Q3 2026 — before the first federation partnership is announced. Federations evaluating us before then can request the working draft under NDA.
The six commitments shown opposite are the operational principles the document will codify. They are stable; they will not change between now and publication.
The six Data Trust commitments
- Anonymization. Athlete names, dates of birth, and identifying metadata stripped from the corpus before training.
- Retention. Corpus contributions retained only while the contributing club maintains Community posture.
- Deletion-on-withdrawal. An individual athlete can request deletion of their contributions; honored within 30 days.
- Audit trail. Every model trained on the corpus carries a manifest of which clubs contributed.
- Third-party review. A named third party with safeguarding credentials reviews the Data Trust annually. Reports published.
- No raw data sale. The corpus is used to train models that ship inside SchoolFish products. It is never sold, licensed, or shared.