Four Malaysian colleagues around a laptop in a meeting room, with a blurred surround

AI meeting etiquette: when not to record and transcribe

Short answer: record and transcribe a meeting only when the people in it have a reason to want a record. Announce the AI note-taker, get consent, and keep it out of sensitive conversations such as performance reviews, discipline, salary talks and client-confidential matters. A transcript outlives the meeting and travels further than the room.

AI note-takers can become part of a team’s meetings before the team has agreed on recording rules. One person connects a tool, and a bot begins joining calls and sending summaries.

A team may have chosen a note-taking tool without discussing every setting it brings. Once people get used to the bot joining, an automatic recording can start to feel like an agreed practice.

AI meeting etiquette includes deciding when a recording serves the people in the room and when it makes the conversation harder. That decision needs more thought than checking whether a tool can transcribe the call.

What the record button changed

Recording a meeting creates a record that can be stored, searched and shared after everyone has left. The people attending need to understand what that means for the conversation they are about to have.

Some assistants can be set to join calendar events automatically. If the owner leaves that setting on, the assistant may appear in a meeting where nobody has discussed recording.

That shift moved the burden. Instead of someone choosing to record, everyone else has to notice the bot and object. Many people will not object in front of their boss, their client or their HR lead.

Etiquette exists for exactly this gap. The person who brings the recording tool carries the responsibility, not the people being recorded.

People speak differently when a bot is listening

People can feel differently about speaking when their words will become a searchable transcript. That is especially important in a conversation where someone needs room to admit uncertainty.

A junior executive might have a concern about a project but hesitate to raise it when her words could be forwarded without the surrounding discussion. An accurate transcript does not remove that concern.

Imagine a project post-mortem where the team needs to admit what went wrong, but participants avoid difficult points because a bot is listening. A colleague asks the assistant, “Summarise what the team learned from this project.” The resulting summary could accurately reflect the polite discussion while missing the concerns nobody felt able to raise.

If the value of a meeting is candour, a permanent record can destroy the thing it was meant to preserve.

Two Malaysian colleagues having a private conversation beside an office counter
Some conversations need room for candour and privacy.

Meetings that should stay unrecorded

These conversations deserve a presumption against an automatic AI note-taker. There may be a justified need for a formal record, but that should be an explicit organisational decision rather than a setting inherited from another meeting:

  • Performance and discipline talks: appraisal discussions, warnings and feedback sessions. People deserve to speak without a searchable file following them for years.
  • Salary and promotion negotiations: both sides may need room to explore possibilities without each opening position being treated as a settled offer. Discuss the purpose of any recording before the conversation starts.
  • Personal and medical matters: health, family circumstances and grievances can involve highly private information. Health information can fall within sensitive personal data under the PDPA; other details need care even where that legal category does not apply.
  • Legal and privileged discussions: conversations with lawyers or about disputes. A careless transcript can surface later in ways nobody intended.
  • Client-confidential sessions: if the client has not explicitly agreed to recording, the answer is no. Their information is not yours to archive.
  • Job interviews: candidates should know whether the interview will be recorded, who will see the record and how it will be used. Do not assume an applicant has agreed to an external AI service processing the conversation.

Where power is unequal or trust is fragile, people may find it difficult to object. The meeting owner needs to consider that before treating silence as agreement.

AI meeting etiquette infographic comparing project updates with sensitive conversations
These are starting points for judgment. Agree the purpose, consent, audience and retention before any recording, including routine meetings.

The legal side in Malaysia

For workplace recordings, a relevant question is whether personal data is being processed in commercial transactions. Malaysia’s Personal Data Protection Commissioner describes that scope for the Personal Data Protection Act 2010.

Purpose, notice, disclosure and retention therefore need attention when a recording includes identifiable people. The team’s PDPA discussion covers related questions about putting personal data into AI services.

The service arrangement matters too. Before adopting a note-taker, the organisation needs to understand where the provider stores recordings, who can access them and what its retention and data-use terms allow.

This is general information, not legal advice. For a policy that fits your organisation, talk to a lawyer who knows Malaysian employment and data law.

Mistakes teams keep making

Most recording problems are not malicious. They come from defaults nobody reviewed:

  • The ghost bot: an AI assistant auto-joins meetings its owner is not even attending, recording rooms with no accountable human present.
  • The forwarded summary: someone mails the AI summary to twenty people without reading it. A transcription error in the second paragraph is now official record.
  • The confident transcript: the summary says a decision was made. Nobody decided anything. The AI compressed a discussion into a conclusion, and the mistake only surfaces weeks later.
  • The infinite archive: recordings pile up with no deletion date, so every offhand comment stays searchable forever.
  • The silent guest: an external partner joins a call and only notices the bot mid-meeting. Even when recording is acceptable, the surprise is what damages trust.

Each of these has the same root: someone treated the tool’s default as the team’s decision.

What good etiquette looks like

You do not need a long policy. You need a few norms everyone actually follows:

  • Announce before, not during: the invite or the first minute says the meeting will be recorded and transcribed. Nobody should discover the bot halfway through.
  • Ask, and mean it: make room for someone to decline without pressure. If someone objects, the meeting owner should be prepared to leave the bot off.
  • Name an owner: one person is responsible for each recording: who gets the summary, where the file lives and when it is deleted.
  • Read before you forward: a human checks every AI summary before it circulates. The sender is responsible for the accuracy of the summary.
  • Set a deletion date: decide how long recordings live. Retaining a transcript indefinitely needs a reason connected to its purpose.

These norms help people understand what will happen to their words. The meeting owner remains responsible for the decision to record even when an assistant handles the recording.

The judgment underneath

A useful starting point is who needs the record and what they will use it for. A project decision log and a sensitive conversation between a manager and an employee can call for different choices.

For a project sync, an agreed record can help absent colleagues understand decisions. The meeting owner still needs to decide whether a full recording is necessary or whether reviewed notes would serve that purpose.

In a conversation with a struggling employee, a manager should consider what record is required and how it will affect candour. An automatic transcript may be inappropriate even if some written documentation is needed.

Deliberate recording means revisiting the choice when the participants or subject changes. Agreement to record a weekly project call should not become permission to record every conversation involving the same people. A trainer can help a team practise these decisions with fictional situations, without asking staff to expose private conversations.

Frequently asked questions

Is it legal to record a work meeting in Malaysia?

Whether a particular recording is lawful depends on its context and use. The PDPA is relevant to personal-data processing in commercial transactions, and employment, contractual or confidentiality obligations may also matter.

Get consent, limit who sees transcripts and delete them on a schedule. For your organisation’s policy, consult a lawyer.

Do I have to tell people an AI is taking notes?

Yes. Beyond any legal question, it is the bare minimum of respect. Announce it in the invite or at the start, and treat an objection as a normal answer, not a problem to argue down.

What if a client asks me to turn the bot off?

Respect the request and leave the bot off. If a written record is needed, discuss an acceptable way to keep notes and agree who will receive them.

Can an AI summary be trusted as official minutes?

Only after a human who attended has read and corrected it. AI summaries compress discussion into tidy conclusions, and they sometimes invent decisions that never happened. The person who sends the summary owns its accuracy.

Should small companies bother with recording rules?

Yes. Even a small team needs shared expectations about who can record, who can see the result and how long it will be kept. The policy should reflect the information involved rather than assume company size makes a recording harmless.

The takeaway

AI note-takers are genuinely useful for the meetings that deserve a record. The skill is knowing which meetings those are.

Before a meeting, consider whether a full transcript is necessary and whether the people attending understand its purpose. Keep the distribution and retention of any record tied to that purpose.

A useful review question is whether the recording helped the people who needed it without exposing a conversation that should have stayed private. The answer may differ from one meeting to the next.

Mike Cheong helps Malaysian teams build the everyday judgment that makes AI tools worth having, including the working norms around recording, verification and trust. Organisations that want their people to use AI with confidence rather than caution-by-avoidance can contact Mike on WhatsApp.

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *