Sovereignty

For organisations working with sensitive data — in healthcare, legal services, financial services, or the public sector — the question of where AI inference happens is not incidental. It affects which data processing obligations apply, which jurisdictional laws govern outputs, and who bears accountability when an inference is wrong. Interface is designed with those questions in mind, even when the current deployment model doesn't resolve them in every possible direction.

In practice, sovereignty at this stage means: transparency about what models are used and under what conditions, clear documentation of where inference is performed, workspace-level isolation so that one organisation's data does not inform another's outputs, and the ability to configure inference frameworks and prompts rather than accepting opaque defaults. It means organisations can understand what is happening to their data well enough to make informed decisions about what they choose to use Interface for.

The longer-term direction this principle points toward is greater infrastructure flexibility — dedicated tenancy, private deployment options, and the ability to bring your own model — as the platform matures. That is where the Sovereign AI research tradition is relevant: not as a description of current architecture but as the intellectual framework shaping where Interface is built toward. The principle commits to that direction; the deployment roadmap is what delivers it.