New AI voice controls and account history: two checks for business owners

CHECK AVAILABILITY, THEN TEST

1. Growthcenter’s underlying platform adds voice to its mobile staff assistant

Confirmed platform announcement · Growthcenter app and account availability unverified

What changed

HighLevel announced two-way voice for Ask AI in its mobile app. Users can speak a question, see a transcript and hear the assistant’s response. The announcement says voice and typed messages share the same conversation history.

The platform says question and approval cards pause the microphone until answered, and users can interrupt a spoken reply. A minimised conversation can remain live, with a floating button to return to it. These are staff-facing Ask AI controls; the announcement is not a release of a new customer-facing telephone receptionist.

Who this affects

Australian owners and staff who already use Growthcenter and want to check whether their mobile app supports this internal assistant. Businesses without that app can ignore the release. A business seeking better inbound call handling should assess its receptionist separately.

Our analysis: what it means for your business

The potential benefit is less typing during a suitable admin task. The practical risk is approving an action after a name, date or instruction was transcribed incorrectly. A clear approval step is useful, but the business still needs to decide what the assistant is permitted to do. There is no measured time saving or booking improvement established here.

Hypothetical example: a Melbourne maintenance owner, parked between jobs, asks for a summary of an approved internal task list. Before authorising any follow-up, the owner checks the transcript and the intended recipient. That is a proposed test, not a demonstrated Growthcenter workflow or a client result.

A voice session that continues when minimised deserves a deliberate finish. In a shared office, avoid speaking customer details where others can hear them, and check whether the microphone is still active before moving to another conversation.

What to do next

1. Confirm the feature is visible in the actual Growthcenter app and account. Record the app version and any permission prompts instead of assuming the underlying platform announcement guarantees access.
2. Start with a non-sensitive internal question while stationary. Compare the spoken request, transcript and answer; check names and dates carefully.
3. For any available action, inspect its scope and approval before allowing a change. Test how to stop voice, return to typing and hand the task to a person. Leave customer-facing receptionist flows unchanged unless they have their own reason and test plan.

REVIEW BUSINESS ACCOUNT ACCESS

2. ChatGPT adds a place to review recent account security activity

Confirmed release note · Individual account availability unverified

What changed

OpenAI’s 25 September release note introduces security history for recent account activity. It includes sign-ins, sign-outs and changes to multi-factor authentication, passkeys and other security settings.

Events include time, location and device information, but OpenAI says some details may be approximate or unavailable. The web path is Settings, then Security and login, then Security history. The release note does not establish that every Australian business account already displays it.

Who this affects

Owners and staff using ChatGPT for business work. Someone who does not use an OpenAI account can safely skip this check; an organisation with an IT administrator should follow its established account-security process.

Our analysis: what it means for your business

Treat this as a useful account-access check. It should not be represented as a complete record of what an assistant changed in a CRM, booking system or other connected service: that broader claim is not established by this announcement. Keep the business’s existing process for reviewing consequential actions.

Hypothetical example: an owner sees an unfamiliar location beside an event. Because location can be approximate, the useful first step is to compare the time and device with their own activity. If it still looks unfamiliar, use the account-security guidance linked from the release note and involve the account administrator. The location alone does not diagnose an intrusion.

What to do next

1. Look for Security history in ChatGPT on the web and review the available recent activity. Note any missing details rather than treating the view as a complete audit.
2. If an event is unfamiliar, compare its context with known device use and follow OpenAI’s linked account-security guidance. Do not share security screenshots containing sensitive account information unnecessarily.
3. Keep access review separate from workflow review: inspect permissions and the destination system’s records when checking an AI-assisted business action.