---
topic: ai-workplace
author: Crashtech Editorial
published: Nov 13, 2026 · read: 5 min
---

Claude Meetings Adds a Visible Participant—and a New Meeting Policy to Set

Claude Meetings can transcribe, answer and build artifacts live; external guests, connected tools and transcript sharing need deliberate controls.

– –

A meeting assistant changes a call even when nobody asks it a question. People may speak differently when a transcript is being created, and notes can outlive the meeting itself. Anthropic’s Claude Meetings beta makes the assistant a visible participant in Google Meet and lets it build artifacts live. That is convenient for a team, but it also means a new person in the room—human or otherwise—needs clear rules.

Colleagues discuss a paper map around a meeting table as a person places an open chair and notepad for a visible assistant participant.
Conceptual illustration of a visible meeting participant; it does not show Claude or a real call.

More than automatic minutes

Anthropic’s November 12 release notes say Claude Meetings is rolling out in beta for Team and Enterprise plans, in Claude and Claude Desktop. In a Google Meet call, Claude appears as a participant, can take notes and answer questions, and can make artifacts such as documents, whiteboards and slides. The help documentation says it saves a transcript, summary and screen descriptions, but not audio. [1] [2]

During a call, anyone on the team can ask Claude to make a document or whiteboard, and Claude attaches the result to the meeting page. The assistant can also use connected tools, such as sending a Slack message, with permission. A “yes” to a first-use prompt applies going forward, so teams should understand which account connection is being authorized before treating a live meeting as a safe place to try tools. [2]

This is different from a note-taker that only produces a recap after the call. Live responses and artifacts can influence the conversation while it is happening. Our look at Asana AI Meetings and reviewable action items offers another example of how a meeting assistant can turn talk into shared work. That can help a team turn a decision into a draft or a whiteboard quickly, but it also creates a chance that an unverified summary or action item will steer the group. Participants should review what Claude writes before treating it as a decision record.

External guests change the sharing boundary

Anthropic’s current help documentation says Claude stays out of outside-guest calls by default on Enterprise, while Team enables outside guests by default and lets an owner turn the setting off. On Enterprise, an owner must allow the feature. If the organization permits external guests, Claude posts a notice in the call chat and includes guest speech in the notes and transcript, but guests do not receive the internal meeting page or recap email. [2]

That creates an asymmetry worth making explicit. An outside participant’s words may be retained in an organization’s transcript even though the guest cannot access the page where that record lives. Meeting organizers should explain the assistant’s role before enabling it, decide whether external calls are appropriate, and avoid putting confidential or personal matters into a call until the organization’s sharing and retention settings are understood.

Anthropic also recommends confirming that everyone is comfortable being transcribed before adding Claude or enabling auto-join. Google Meet offers administrative controls for explicit participant consent for notes, transcripts and recordings, but that setting is configurable and does not automatically govern every organization. Teams should use their actual administrator settings and follow the policies that apply to their calls. This is product guidance, not a statement about the law in every jurisdiction. [2] [3]

Set controls before auto-join

On Team, Claude Meetings is on by default, although an owner can turn it off. On Enterprise, the owner enables it for all users or selected roles. Auto-join is configured by each member; owners cannot set it for them. Organizations can also control whether external guests are allowed, whether members’ connectors can be used, and which custom roles may use Meetings. [2]

Before enabling the feature broadly, write a short internal policy that answers four practical questions: which meetings may include Claude, how participants are notified, whether external guests are permitted, and who can connect tools. Offer a way to pause or remove the assistant during a call, and identify who can access and share the resulting notes. For sensitive meetings, keeping auto-join off may be the simplest rule.

Then test the lifecycle with a non-sensitive meeting. Confirm the participant is visibly present, the notice appears, the transcript and artifacts land in the expected internal location, and the recap is shared only with intended colleagues. Test what happens when an external guest joins late or when someone declines. Anthropic documents how Claude behaves in these cases; teams should confirm that behavior with their configured account and current rollout. [2]

A beta that can create durable records

The notes and transcripts are included in Team and Enterprise plans, according to Anthropic. Documents and whiteboards created in meetings are free through February 10, 2027; use of connected tools counts toward the user’s plan and can incur organization charges. Beta availability is limited for some configurations, including zero data retention, HIPAA-ready settings and customer-managed encryption keys. These limits make it important to confirm eligibility and costs before a team plans a workflow around the feature. [2]

A meeting transcript can contain more context than an official set of minutes. Decide whether these records belong in the same retention and access system as other business documents, and whether they should be shared outside the invite list. A privacy framework such as NIST’s can help teams map data, access and risk without assuming that a meeting tool’s product controls settle every internal policy question. Our enterprise AI data-custody guide offers a broader checklist for retained work records. [4]

Make the assistant welcome before it joins

Claude Meetings could reduce the friction between a discussion and its first useful artifact. The tradeoff is that its output becomes part of the room’s information flow and a lasting record. A visible tile is helpful, but visibility alone is not consent, policy or access control.

Start with a narrow group, internal meetings and auto-join disabled. Explain what is captured, check the external-guest setting, and review who can see the resulting page. Expand only after participants and administrators understand the feature’s behavior. A good meeting assistant should make the work easier while leaving no uncertainty about who heard the discussion and where its words went.

Advertisement

Frequently asked questions

What does Claude Meetings do?

Anthropic says Claude can join a Google Meet call as a visible participant, take notes, answer questions and create artifacts such as documents or whiteboards. It saves a transcript, summary and screen descriptions, but not audio.

Can Claude Meetings join calls with external guests?

Team plans allow this by default and owners can turn it off. Enterprise plans require an owner to allow external guests. When enabled, guest speech is included in the transcript and notes, while guests do not receive the internal meeting page or recap email.

Can Claude use connected tools during a meeting?

Yes, when members have connected tools enabled. Anthropic says Claude asks for permission the first time a tool is used, and a yes applies going forward. Organizations should set connector and approval policies before enabling the feature.

Sources & further reading

/* Comments */