Privacy and data boundary

Audit Consent in Local and Hybrid AI

Trace consent state from user choice through routing, payload selection, revocation, and disclosure.

Editorial review pending · Not indexed · Fact IDs: ARCH-02, PRIVACY-02

Direct answer

Trace consent state from the user's initial choice through each subsequent routing step, payload selection, revocation handling, and disclosure. Do not infer anything beyond this specific flow when reviewing the audit.

Evidence to collect

The claim becomes reviewable only when the following evidence is attached to the same artifact and test run:

  • Consent-state record
  • Route and payload capture
  • Revocation test

A privacy claim needs a data inventory and traffic evidence. A local-model label by itself is not proof.

Implementation workflow

Link together the consent-state record, the route and payload capture, and the revocation test by recording each item in sequence and comparing them side by side to verify that consent handling remained consistent from selection through to final disclosure.

  1. Inventory every content and metadata type.
  2. Enumerate every destination and trigger.
  3. Map user action to code path and request fields.
  4. Verify the core task in airplane mode.
  5. Capture traffic for setup, inference, diagnostics, and export.
  6. Update public disclosure in the same release as behavior.

Failure patterns to prevent

  • Using stale consent
  • Transferring before the decision

Avoid relying on stale consent data, and do not allow transfer of information before the consent decision has been finalized. Limit all release wording strictly to the evidence that has been recorded, as associated with ARCH-02 and PRIVACY-02.

Minimum reproducibility record

LayerRecord
DeviceManufacturer, model, chipset, RAM class, operating-system build
SoftwareApplication version and git commit
ModelFamily, variant, revision, format, file length, hash, quantization
RuntimeName, revision, requested backend, observed backend evidence
WorkloadFixture revision, input hash, prompt hash, output policy
OutcomeCompleted, failed, cancelled, fallback, and privacy-safe diagnostics

Release checklist

  • ☐ The primary query is answered without an unsupported number.
  • ☐ Every artifact and runtime is pinned.
  • ☐ The representative task and failure policy are explicit.
  • ☐ Lifecycle, cancellation, cleanup, and fallback are tested.
  • ☐ User-content and network boundaries are documented.
  • ☐ Result wording applies only to the recorded configuration.
  • ☐ The page links to raw method or evidence when results are added.

Sources and related evidence

Chinese deployment and troubleshooting content is organized in the .奇连 AI 端侧专题.