An answer's provenance should include where it ran

When an AI system gives an answer, we increasingly care which model produced it. I think that record is incomplete.

Where the model ran can matter just as much. Was the request processed on a device the person controls, on nearby shared hardware, or in a remote service? What version ran there? What information was allowed to leave with the request?

Location is not merely an infrastructure detail if it changes privacy, latency, cost, or trust. It is part of the answer’s provenance. The route should be something the system can explain, even if the user does not need to approve every step.

The goal is visibility without interruption: a record beside the answer, not another confirmation before every harmless request.