Claude is now listening to your Google Meet calls, and consent is your problem

Anthropic’s launch video shows a bot building a whiteboard while people talk. It leaves out the settings that decide who may be in the room, who has agreed to be transcribed, and what happens when the free period ends on 10 February 2027. For Team admins, those gaps matter more than the demo.

AI meeting bots no longer wait for the call to end. Anthropic’s version writes a plan, builds a timeline and answers questions while people are still talking, and the first control that decides where it may sit, whether outside guests can be on the call, is described two ways by the company that made it. Anthropic’s announcement and its Help Center disagree on the Team-plan default, which decides whether a bot you invited will also listen to a client.

The example is Claude Meetings, shown in a promotional video that Anthropic’s own Claude channel published on 12 November 2026, the day the beta began rolling out to Team and Enterprise plans on Google Meet. The video is a maker’s advertisement and it shows the best minute of a call. This article checks its claims against Anthropic’s documentation, independent reporting and the legal commentary that already exists on meeting notetakers.

A promo that shows the best minute of a call

The video’s description makes five claims. Claude joins as a visible participant, answers a question asked aloud or in chat, starts a doc on request, lets the team watch and correct what it builds, and leaves notes, action items and a transcript on a page in Claude after the call. Anthropic’s announcement confirms each one and adds detail the video leaves out. By default Claude answers in writing, it shows in the participant list and posts a notice in the call chat, and it keeps a transcript, a summary and descriptions of shared screens, not audio.

The transcript of the opening seconds carries greetings, one spoken request to start a whiteboard, and music. It carries no spoken reply, which fits the documented default of written answers. Neither the description nor that transcript mentions settings, guests, consent, price or limits. That is normal for an advertisement, and it is the reason the rest of this article exists.

The dates also need a note. Anthropic’s post of 10 November called Meetings coming soon, TechCrunch covered the announcement that day, and the beta rollout began on 12 November, the date on Anthropic’s launch article and on the video.

The guest rule that two Anthropic pages state differently

Anthropic’s launch article says Claude is set to stay out of calls when someone from outside the organisation is invited, and to leave if an outsider joins midway. An admin can allow it. TechCrunch, IT Brief and Dataconomy repeat that version, and so does the AI Codex guide for admins.

The Help Center page for Meetings says something else. For Team plans it states that Claude can join calls with outside guests by default and that an owner can turn this off. For Enterprise plans an owner must allow it first. The owner settings on the same page list the defaults side by side: Team allows external guests, Enterprise is internal only. The page carries an update stamp later than the announcement, and consultant Justin McKelvey flagged the mismatch on 14 November, advising Team owners to set the rule deliberately instead of guessing which page their account follows.

Nothing in the sources settles which page is right. The Help Center is more specific and was edited later, so it is the better guide to how Team accounts behave today, but that is an inference and not something Anthropic has said. When guests are allowed, the documented behaviour is narrow. Claude listens, posts its notice in the chat without replying there, includes what guests say in the notes and transcript, and does not give guests the meeting page or the recap email.

Consent sits with the person who adds the bot

Anthropic’s Help Center tells users to “make sure everyone on your call is okay with being transcribed” before they add Claude or turn on auto-join. Auto-join can cover all meetings or all except one-on-ones, and an owner cannot set it for a member. The product’s safeguards are about visibility. Claude shows in the participant list, posts a notice in the chat, can be removed from the meeting page, and when it has to ask to enter, the host can decline. None of that is a consent step. Answers Claude posts in the chat are visible to everyone on the call, and it uses a person’s connected tools only with that person’s permission.

Otter.ai litigation sets the legal background

The legal background is unsettled. According to Littler employment lawyers, quoted by HR Executive, federal wiretap law and most state laws need one party’s consent, but about a dozen states require every participant to agree to recording or interception. The best-known test is the consolidated Otter.ai privacy case in the Northern District of California, where plaintiffs allege that a notetaker recorded private conversations without all participants’ consent. When that coverage ran in April 2026 no substantive ruling had been issued, so the claims remain allegations, not findings. Lawyers at Robinson & Cole add that some complaints say the vendor put the burden of getting permission on the account holder, and that a tool which only transcribes may still fall inside privacy statutes.

No court has examined Claude Meetings. Its visible notice separates it from the complaint that a bot joined silently, but whether a chat notice and a host’s say-so satisfy an all-party law is the open question, and these sources do not answer it. That reading is an inference, not a legal finding. A company with staff or customers in all-party states should ask its own counsel.

Google Meet only, and closed to regulated setups

Google Meet is the only supported platform. Anthropic says other video platforms are coming soon and gives no date. McKelvey notes that Granola and Microsoft’s Copilot already work on Zoom or Teams calls, so teams on those platforms cannot use Meetings yet. The beta also excludes organisations with zero data retention, a HIPAA-ready configuration or customer-managed encryption keys, according to both Anthropic pages. Enterprise owners must switch Meetings on, for everyone or by custom role, while Team owners find it already on.

The competitive context explains the speed. TechCrunch ties the launch to OpenAI’s workplace tools, unveiled a few weeks earlier, and to Meta’s Muse agent, all aimed at taking on Microsoft and Google. IT Brief reads it as part of a wider push to put assistants inside workplace communication tools instead of a separate chat window. Both are journalists’ framing and not Anthropic statements, and neither changes what a Team owner has to decide this week.

Free until 10 February 2027, then no stated price

During the beta, notes and transcripts are included on Team and Enterprise plans. So are Claude’s answers in calls, unless an answer uses a connected tool, in which case the usage counts against the plan. Docs, whiteboards and other artifacts built in a call are free through 10 February 2027, according to the Help Center. Dataconomy reports that Anthropic has not disclosed what those artifacts will cost afterwards, and McKelvey tells readers to expect them to count like other artifacts. That is his expectation, not Anthropic’s statement.

The seat is the standing cost. Dataconomy, citing Anthropic’s pricing page, lists Team standard seats at $20 per user per month on annual billing or $25 monthly, and Enterprise at $20 per seat on annual billing with a 20-seat minimum and usage billed at API rates. McKelvey gives the same Team figures for teams of 2 to 150 people. The beta itself has no announced end date. A team that builds its Monday routine around live docs will reach 10 February 2027 without a published price for them.

Settings to fix before the next client call

For a Team-plan owner, the evidence supports a plain order of work. Open Organization settings, then Products, then Claude Meetings (beta), and set the outside-guest option on purpose, because the two Anthropic pages cannot both describe your account. Decide on auto-join, which AI Codex points out makes the consent check easy to skip. Tell guests before a call that a bot will transcribe; the step is cheap and sidesteps the open legal question. Keep regulated work out until the exclusions lift, and put 10 February 2027 in the calendar.

The video’s promise holds up. The product exists, works in Google Meet and does what the description says, according to Anthropic’s pages and the independent coverage. What the video omits decides whether a team can use it safely, and the guest default is the item to resolve first. Three developments would change this reading: Anthropic reconciling its two pages, a ruling in the Otter.ai case on notetaker consent, and support for platforms beyond Google Meet.

Author:
Jan Bielik
CEO & Founder of Webiano Digital & Marketing Agency

This article is an original analysis supported by the sources cited below

Introducing Claude Meetings
The promotional video itself, published on Anthropic’s Claude channel on 12 November 2026, with the description that states the product claims.

Introducing Claude Meetings, now in beta
Anthropic’s launch article: how the bot joins, what it saves, the outside-guest default as Anthropic first stated it, beta limits and pricing.

Use Claude Meetings
The Help Center page with the Team and Enterprise owner defaults, the consent advice and the 10 February 2027 date for free artifacts.

More ways to work with Claude: Sheets, Whiteboards, and Interactive Connectors
Anthropic’s 10 November post, which still called Meetings coming soon two days before the beta.

Anthropic launches Claude Meetings to turn workplace discussions into plans and tasks
Independent report from 10 November on the launch, the guest rule as announced and the OpenAI and Meta competition.

Anthropic launches Claude Meetings for live video calls
Independent summary of the controls, exclusions and beta pricing, and the view of a wider push into workplace communication tools.

Anthropic brings Claude Meetings to Google Meet in beta
Source for the Team and Enterprise seat prices as listed on Anthropic’s pricing page and for the missing price after 10 February 2027.

Claude Meetings (Beta): Claude Joins Your Google Meet, Takes the Notes, and Builds the Doc Live (2026)
A consultant’s review that spotted the guest-default mismatch between Anthropic’s two pages and compared Google Meet-only support with Zoom and Teams tools.

Claude Meetings: Claude joins your Google Meet calls. The four settings to decide before anyone invites it
Admin guidance on external calls, auto-join and connectors, repeating the announcement’s version of the guest default.

When AI Notetakers Take the Stand- The Legal Risks Lurking in Your Virtual Meetings
Robinson & Cole analysis of the Otter.ai complaints and the consent risk of notetakers that only transcribe.

The notetakers are ready to start the meeting, but will they always be so welcome?
Littler’s count of about a dozen all-party consent states and the status of the consolidated Otter.ai case in April 2026.

Citing this article? Brief excerpts are welcome. Please credit Webiano.digital, name the author where stated, and include a link to https://webiano.digital and to this original article. Full or substantial republication requires prior written permission. Read our Copyright and Content Use Policy.

This article was prepared with the assistance of artificial intelligence tools. The content underwent expert human review, and Webiano Digital & Marketing Agency assumes editorial responsibility for its final version and publication.