BURKI GPT LIVE: EVERY-CALL RECORDING DISCLOSURE
Free configuration worksheet | Prepared October 25, 2026

This resource is free. Production voice, recording and telephone usage may
cost money. It does not authorise a paid call, a settings change or a release.
Current scope: every-call disclosure before GPT Live, using the separate
Announcement voice. First-call-only suppression is not established here.

1. NAME THE PROCESS BEFORE OPENING THE EDITOR

Business / workspace:
Assistant ID and name:
Operator responsible for recording policy:
Jurisdiction and approved recording process reference:
Script owner and approval date:
Script version:
Which calls should be recorded:
Why audio is retained:
Who may access it:
Approved contact option if someone objects:
Who verified that option and when:

Approved disclosure text:
[Replace this field with the business's approved wording. Do not leave a
template sentence in a customer route without checking its truth and scope.]

ILLUSTRATIVE PRACTICE SCRIPT, NOT UNIVERSAL CONSENT WORDING:
This call is being recorded for service review. If you do not want to continue
on a recorded call, please tell us so we can explain the office contact option.

Check whether "is being recorded" accurately describes the verified route.
Do not invent an unrecorded alternative, automatic deletion or a human transfer.

2. CONFIGURATION AND ELIGIBILITY

Record evidence, not just a check mark:

Engine: GPT Live / other / not checked
Publish -> Published version state:
  published / unpublished / legacy live settings / unknown
Current settings revision or evidence reference:
Intended publication revision:
Recording enabled:
Auto-save configuration checked:
Workspace retention/policy permits recording:
Browser recording consent, if applicable, bound to current user/workspace:
Funded recording eligibility:
Actual admission blockers:
Cost/allowance reviewed at:
Who separately authorised any test, and its scope:

Do not turn on recording to fill out this worksheet. Review existing settings
first. A transcript does not establish stored audio. Free practice does not
record calls. A disclosure switch does not override workspace retention policy.

3. CURRENT EDITOR PATH FOR AN AUTHORISED CHANGE

Saved assistant -> Publish -> Safety settings
  Enable recording
  Recording disclosure, exposed when recording is enabled
  Disclosure message, exposed when disclosure is enabled
  Frequency -> Every call (recommended)

The last two controls are conditional. Their source labels do not mean they
were visible in a saved assistant whose disclosure switch was off. Do not use
First call only per caller as a promise of suppression in the native path.

GPT Live assistant -> Voice -> Announcement voice
  Provider:
  Model:
  Voice:
  Credential or managed-funding readiness:
  Separate speech quote / rate evidence:
  Available funding checked at:

This voice is separate from GPT Live's conversational voice. Do not change a
saved assistant's engine merely to inspect the panel. Do not infer provider
readiness from a selected voice name.

4. KEEP THE GREETING AND OBJECTION RULE SEPARATE

First greeting:
[Approved greeting guidance for the conversation after the announcement.]

The GPT Live greeting may be paraphrased. The disclosure is separately
configured speech. Do not duplicate it in the greeting to conceal a failed
announcement path.

Adaptable instruction for the fictional Juniper Repairs Reception example:

If a caller objects to recording, acknowledge the objection and explain the
approved office contact option from the business facts. Do not claim you paused
recording, deleted audio, changed consent or connected a person unless an actual
supported action establishes that result. Do not argue with the caller or keep
collecting repair details after they decline this recorded conversation.

This instruction does not install a recording-control or transfer action.

5. VERSION REVIEW

Read Publish -> Published version before changing an existing assistant.
Published assistant: saved edits remain drafts until Publish saved changes.
Legacy live settings warning: saving changes connected calls.
New assistant: Create assistant, Publish assistant and telephone connection
are separate steps. Credentials and workspace policies remain live dependencies.

Configuration reviewer:
Version reviewed at:
Release decision and authorisation:
Rollback reference:
Native worker revision actually established: UNKNOWN until verified

6. ACCEPTANCE CASES, NOT PERFORMED RESULTS

All observed results begin UNTESTED. Use an approved test process separately.

A. First recorded call
Expected: approved every-call announcement precedes the GPT Live conversation.
Observed: UNTESTED
Configuration / route / time / evidence:

B. Repeat caller
Expected: the every-call announcement still applies; no suppression promise.
Observed: UNTESTED
Evidence:

C. Caller objects to recording
Expected: approved contact process; no invented pause, deletion or transfer.
Observed: UNTESTED
Evidence:

D. Missing announcement credentials or quote
Expected: admission cannot proceed as though separate speech were ready.
Observed: UNTESTED
Evidence:

E. Announcement playback or cleanup is uncertain
Expected native behaviour: stop startup rather than silently begin GPT Live.
Observed: UNTESTED
Running worker compatibility / error / evidence:

F. Audio exists in history
Expected: review whether it actually plays; a listed file is not sufficient.
Observed: UNTESTED
Stored-audio evidence:
Actual caller-end listening evidence, if required and authorised:

7. CLOSE ONLY THE CHECKS YOU HAVE ESTABLISHED

Script approved: UNKNOWN
Settings saved: UNKNOWN
Publication version reviewed: UNKNOWN
Provider admission: UNKNOWN
Separate announcement playback stage: UNKNOWN
Caller heard complete announcement: UNKNOWN
Recording plays: UNKNOWN
Business recording process accepted: UNKNOWN

A successful save does not fill the remaining fields. Preserve the actual error
and unresolved checks. Do not remove a disclosure just to make a test pass.

Original article:
https://burki.dev/blog/burki-gpt-live-recording-disclosure
Commercial context:
https://burki.dev/use-cases/ai-receptionist
https://burki.dev/pricing