AI Principles

The principles

• The firm's data is never used to train a model, ours or a provider's.
• A provider keeps nothing it is sent beyond the time needed to answer, and never for its own purposes.
• Every AI feature is off until the firm turns it on, and the firm chooses which provider, where one choice is offered.
• What AI drafts is marked as such and is reviewed by a person before it is relied on or sent; it never acts alone on a client's matter.
• Information walls and permissions apply to AI as to people: it sees only what the person using it may see.
• AI use is recorded in the audit trail like any other action.

Before any feature launches

We publish here the provider, what it is sent, where it processes data and the agreement that binds it, and add it to the Subprocessors page with 30 days' notice.