Meeting consent and transparency with Granola: Chat notification, video watermark and admin enforcement

TL;DR: Enterprise conversations need clear disclosure and clean data handling. Tell participants at the start that you're using Granola to transcribe the call, and turn on Granola's transparency features: an automated chat message and a watermark on your video, available on Zoom, Google Meet and Microsoft Teams. Any user can turn them on, and Enterprise admins can enforce them across a workspace. Granola runs on your laptop or phone and uses your device's audio to generate notes, and it doesn't keep meeting audio once the transcript is done, so recall stays accurate and security reviews stay straightforward.

Across sales calls, executive briefings, client discussions, and sensitive internal meetings, the thing that most often degrades a record of the conversation is a participant trying to type while also listening. Telling people up front that you're using an AI notetaker to transcribe the call, and then being able to stay in the conversation rather than in your notes, is what produces an accurate record that everyone in the room agreed to.

This guide covers how to build a repeatable AI notetaker consent workflow: what to tell people and when, how Granola's chat message and video watermark let participants know it's transcribing, what to do when someone says no, and what to tell participants about their data. It covers how consent works in practice, not the law. For the rules where you and your participants are, check with your legal team.

Why consent matters when you use an AI notetaker

Everyone in the meeting should know

Everyone in a conversation has a stake in how it's captured. Telling people you're using an AI notetaker, and giving them a clear chance to say no, is the standard Granola asks every user to follow, whether the meeting is on a video call, in person or on the phone.

Tools differ in how they capture a meeting: some join the call as a participant, and some, like Granola, run on your laptop or phone and use your device's audio to generate notes. Whichever you use, tell the other participants at the start of the call that you're transcribing. It takes a sentence, and it's the standard everyone in the room should expect. In a Zoom, Google Meet or Microsoft Teams call, Granola can help notify participants with a chat message or a watermark on your video.

Consent and confidentiality in enterprise meetings

Consent is a trust signal. Informed consent means people know four things before the conversation starts: that an AI notetaker is transcribing, what it keeps, who can see the notes, and how long they're kept. A vague "this call may be recorded" may not give participants enough information to understand what is happening.

What teams can control is the infrastructure: choosing a transcription approach that supports automatic participant notification, deletes audio after transcription and has clear, documented retention and deletion policies. Getting the architecture right, and telling participants plainly that you're transcribing, is what earns trust in the process.

How Granola captures a meeting

How Granola captures audio

Granola runs on your laptop or phone. It captures audio directly from your device, transcribes it (in real time on Mac and Windows), and then deletes the audio. Granola doesn't store meeting audio. Because it works from your device rather than inside the meeting app, letting people know is up to you and its transparency features. The AI notepad sits on your desktop alongside your video call, and Granola offers built-in ways, like a video watermark or an automatic chat message, to let other people in the meeting know it's active.

Stay present in every meeting

When you are not typing notes, you focus entirely on the conversation. Whether you are a sales rep reading a prospect's hesitation, an executive navigating a sensitive negotiation, or a manager coaching a direct report, presence is the variable that determines meeting quality. You maintain eye contact, catch the moment the room shifts, and respond in real time rather than catching up later. Granola's AI-enhanced notes workflow is built around this dynamic: you jot what matters during the call, and the AI fills in context from the transcript afterward. Your rough notes guide the enhancement, so the final output reflects your priorities rather than a generic summary.

How Granola lets participants know it's transcribing

Granola gives you two transparency features to make it clear that notes are being captured: an automated chat message and the Granola Watermark. You can turn on one or both, just for yourself or across your whole team.

Automated chat message

When Granola starts transcribing, it automatically posts a message in the meeting chat to let participants know. It's automatic, so you don't have to remember a manual step, and everyone in the chat can see it.

Granola Watermark

The Granola Watermark shows a visible indicator on your video that Granola is transcribing. It stays on throughout the call, so people who join late can see it too, and it turns off when you stop transcribing.

Where the features work and who can turn them on

Enforce notice across your workspace on Enterprise

On the Enterprise plan, admins choose which transparency features to use across the workspace and enforce them from central controls, so the same notice applies in every meeting without relying on each person to switch it on. Read more about Granola's admin controls.

Add a meeting-level notice in Zoom or Microsoft Teams

Some organizations also set a notice in the video-conferencing platform itself. Zoom account owners and admins on eligible plans can show a custom disclaimer when participants start or join a meeting, and Microsoft Teams admins can customize the recording and transcription notification message. Granola doesn't configure these settings for you, but they work alongside its chat message and watermark. Granola's guide to letting people know you're using Granola links to the Zoom and Teams instructions.

In-person meetings and phone calls

The chat message and watermark work in video calls. In an in-person meeting or a phone call on your phone, tell people at the start that you're using Granola to transcribe the conversation and ask if they're comfortable with it.

How to get consent for AI notetaking, step by step

When to disclose AI-assisted note-taking

Disclosure should happen before transcription begins, not after. Inform participants in the meeting invitation that AI-assisted transcription will be used, then confirm verbally at the start of the call. The disclosure timing table below makes this repeatable:

A one-line calendar note can be as simple as: "I'll use Granola, an AI notepad, to transcribe this call so I can focus on the conversation. Let me know if you'd prefer I take notes by hand."

Granola's in-meeting notice feature sends a message to meeting chat when Granola starts transcribing, so participants are notified without requiring you to remember a manual step. Granola's getting-consent guide covers the recommended disclosure approach for different meeting types.

Verbal consent scripts for enterprise conversations

Best practice separates consent for the meeting itself from consent for transcription. This applies whether you're running a sales discovery call, an executive briefing, a client check-in, or a research interview. Asking permission to transcribe as a distinct step keeps the conversation straightforward and gives participants a clear moment to opt out. Here are three scripts calibrated for different professional contexts:

• Sales and discovery calls: "Before we jump in, I want to let you know I'm using an AI notepad to help me capture our conversation accurately so I can stay focused on you rather than my notes. The tool transcribes in real time to help me follow up accurately. Are you comfortable with that?"
• Client meetings and executive conversations: "I'll be using an AI-assisted transcription tool today so I can give our discussion my full attention and still capture the details accurately. The audio is transcribed and then deleted, so no audio recording is kept. If you'd prefer I take manual notes instead, just let me know. Are you comfortable proceeding?"
• Internal team and cross-functional sessions: "Are you comfortable with me using an AI notepad to capture our discussion? It transcribes in real time so I can focus on the conversation."

Best practice for verbal consent is to capture the participant's affirmation in the transcript so consent is documented in the same record as the conversation. Ask a simple yes or no question before the substantive discussion begins, and let the participant's spoken response appear in the notes.

Storing participant consent safely

Verbal consent documented in a transcript is a starting point, not a complete record. Document the participant's agreement in your research repository, CRM, or project management system. Note the date, participant identifier, the tool disclosed, and the opt-in confirmation. Because the spoken yes sits in the transcript, you can find it later by asking Granola Chat.

What to do when someone says no

When a participant declines transcription, stop Granola, acknowledge the choice without friction, and proceed with manual notes. Keep a template ready in a second window. If the conversation turns sensitive partway through, pause transcription and restart when everyone is ready. If something gets transcribed that shouldn't have been, you can delete that part of the transcript, regenerate the note to make sure there's no mention of what you just deleted, and keep the rest. Document the opt-out in the same place you document consent, so your archive reflects the complete picture of how each session was conducted.

What to tell participants about their data

What happens to the audio

Stored audio files create risk on multiple fronts: they are high-value targets for attackers, they accumulate indefinitely, and they require extensive controls to protect participant privacy.

Granola for Mac and Windows transcribes in real time and doesn't store the audio, only the transcript. Granola on mobile caches the audio temporarily during the meeting and deletes it once transcription is complete. No audio recording exists for third parties to use for model training.

Ensuring privacy with data deletion requests

Participants can request deletion of their personal data at any point. Because Granola does not store audio, deletion requests are simpler to fulfill. The transcript and enhanced notes are the primary data artifacts. Granola's transcript auto-deletion feature lets organizations on the Enterprise plan configure org-wide auto-deletion periods so transcript data is removed on a defined schedule without manual intervention for every record.

Criteria for data retention and deletion

Enterprise data lifecycle policies for meeting transcripts typically address:

• Retention purpose: The reason data is kept, documented in advance.
• Retention period: How long after the meeting the transcript will be retained.
• Deletion trigger: What event or schedule initiates deletion.
• Access controls: Which team members can view, query, or share transcripts.
• Deletion verification: How the organization confirms that deletion has occurred.

How Granola handles data retention

Granola's data handling aligns with how teams actually think about participant data. Meeting audio isn't kept, and your notes are private by default: only you can see them unless you share a link or add them to a team folder. Transcripts can be set to auto-delete on a defined schedule through org-wide configuration, so data is removed without manual intervention for every record. Granola also contractually prohibits third-party AI providers from training on your data. For retention windows, redaction and deletion in regulated teams, read Granola for regulated teams.

How Granola handles security and model training

SOC 2 Type 2 and security

Granola holds SOC 2 Type 2 certification. Granola's security page covers the full architecture, including data processing, deletion, and compliance documentation. For SOC 2 and GDPR in more depth, read AI notetaker privacy and compliance: SOC 2 and GDPR.

No third-party AI training on your data

Granola contractually prohibits third-party AI providers from training on your data. All users can opt out of Granola's own anonymized training in settings. Enterprise plans extend this to the entire organization by default: the model training opt-out is on for every user in the org without requiring individual action.

Setting up a consent workflow for your team

Pre-meeting consent checklist

Use this checklist before every sensitive conversation involving AI-assisted transcription:

• Disclose in the invitation: Include a one-line note in the calendar invite stating AI-assisted transcription will be used.
• Turn on Granola's transparency features: Switch on the chat message, the watermark or both, or check with your admin whether they're already enforced for your workspace.
• Prepare your opt-out plan: Have a note-taking template ready in case a participant declines.
• Check enterprise participant restrictions: For B2B meetings, confirm transcript data stays within your organization.
• Confirm your retention policy: Know how long the transcript will be retained and be able to tell the participant if asked.
• Open consent at the start of the call: Run your verbal consent script before the session begins.
• Document the agreement: Note the participant's consent in your research repository immediately after the call.

Controlling who sees insights

Preserve institutional memory across meetings while controlling who can see what. Shared folders can be organized by project, team, or initiative, with members seeing only folders they've been invited to. Folder-level queries let authorized members surface patterns across all conversations without sharing individual transcripts with the entire organization. On Enterprise, the audit API lets your security tools pull workspace audit events. Granola's enterprise features overview covers SSO, SCIM, granular user access control, consent management, and org-wide auto-deletion periods in full detail.

Maintaining ethical consent trails

Build a repeatable documentation habit: after every meeting, record the attendee details, consent status, tool disclosed, retention period, and any opt-outs. Store this record alongside the meeting notes, not in a separate system that might not survive a team transition.

When someone leaves and a new team member takes over, the consent trail shows exactly how conversations were documented, what participants were told, and what obligations the organization still holds. Any team managing sensitive conversations, whether in research, recruiting, or client work, carries continuity forward without gaps or ambiguity.

Granola's consistent weekly retention reflects something simpler than habit: the notes get better the more you use them. Every meeting adds to an archive that stays accessible and searchable across the whole organization.

Try Granola for free to see how accurate your meeting notes can be when you're not typing them yourself. Download the Mac, Windows, iPhone or Android app, connect your calendar, turn on the chat message or watermark, and run your next meeting.

FAQs

Do people in a meeting see that I'm using Granola?

Granola can notify meeting participants that it's transcribing. Turn on its automated chat message, the Granola Watermark or both. These features work with Zoom, Google Meet and Microsoft Teams and are available on all plans. Enterprise admins can enforce them across the workspace. Because a notification may be missed, Granola recommends giving participants a verbal heads-up too.

Does Granola require the other person to consent?

Granola encourages users to get consent: its guidance is to always let participants know when you use AI notetaking, whether you're on a video call, in person or on the phone. Whether the law also requires consent depends on where you and the other participants are, so check with your legal team.

Does the chat message or watermark count as consent?

They give notice, not agreement. They tell people Granola is transcribing, but they don't confirm that anyone agreed. Pair them with a yes or no question at the start of the call so the answer appears in the transcript, and document their response according to your organisation's consent process.

What happens if someone doesn't want Granola on?

Stop transcription and take notes by hand. If you've already started, you can delete the part of the transcript that shouldn't be kept and keep the rest.

What is informed consent for AI notetakers?

Informed consent means every participant is told before the session begins that AI transcription will be used, understands how data will be stored and accessed, and actively agrees to proceed. Documenting verbal consent at the start of the call in your records is a simple way to keep that agreement on file.

What is an enterprise data deletion policy for AI meeting tools?

An enterprise data deletion policy sets out how long meeting data is retained and when it must be deleted. Granola does not store audio, only the transcript and your notes. Transcripts can be configured for org-wide auto-deletion on a defined schedule through Enterprise admin controls, and participants can request deletion of their data at any time.

Key terms glossary

Transparency features: Granola's automated chat message and video watermark, which notifies everyone in a Zoom, Google Meet or Microsoft Teams call that Granola is transcribing.

Granola Watermark: A visible indicator on your video that shows Granola is transcribing. It stays on for the whole call and turns off when transcription stops.

Device audio: Audio captured directly from your device's microphone and, on desktop, system audio. Granola uses this audio to transcribe conversations from any meeting app or in person.

Meeting archive: A centralized, searchable collection of meeting transcripts, notes, and synthesized insights that enables teams to query past conversations and build institutional memory across all meeting types over time.

AI training opt-out: A setting that prevents a vendor or its AI providers from using your data to train or improve machine learning models, available as an org-wide default on Granola's Enterprise plan.