Only the learning leaves. Here's what that means.
- Is my data safe?
- It never leaves. Training comes to you.
- What do I earn?
- A share of metered usage, quarterly, on a published formula.
- Can I leave?
- Any time. Your influence stops accruing to new training rounds.
Your data stays put
OpenLegal never ingests, stores or transmits your documents. Training executes within your controlled environment — your cloud tenancy, your hardware, or a jointly attested secure enclave. Your choice.
Learning is aggregated, not text
What returns to the network are model updates — mathematical adjustments — passed through privacy defences designed to prevent source text being reconstructed. We publish our methods and invite independent audit.
Contribution is metered
Every contributor's influence on the model is tracked, and usage of the model is metered. The mapping from influence and usage to payment is a published formula, not a black box.
Exit is real
Leave any time. Your influence stops accruing to new training rounds, and no document was ever ours to keep — because we never had it.
Where an assurance is still being audited, we say so on this page rather than in an adjective. The audit ships before the claim does.
Being a customer of a product built on OpenLegal contributes nothing to the model.
Applications use the model. They never feed it. Contributing is a separate choice, made directly with OpenLegal under its own agreement. The two never mix. FinePrint, an independent company, is OpenLegal's first paying customer — and its customers are not contributors.