How to stop AI note takers joining your meetings (Teams, Zoom and Google Meet)

In the Teams admin center, go to Meetings, then Meeting policies, then Meeting join & lobby, and open "Manage external bots and their access to meetings". The options are "Do not detect bots" (the PowerShell value AllowBots) and "When detected, require approval before joining" (RequireApprovalWhenDetected, the default). Keep the policy on the approval setting so detected bots wait in the lobby instead of joining silently. You can read and set the value in PowerShell:

Note that Microsoft Learn's own page on external bots shows PowerShell examples that call Set-CsTeamsEventsPolicy instead of Set-CsTeamsMeetingPolicy — a known documentation error. The correct cmdlet is Set-CsTeamsMeetingPolicy, as shown above.

Connect-MicrosoftTeams
Get-CsTeamsMeetingPolicy -Identity Global | Select-Object ExternalBotAccessMode
Set-CsTeamsMeetingPolicy -Identity Global -ExternalBotAccessMode RequireApprovalWhenDetected

Microsoft rolled out a block option for detected external bots (message centre announcement MC1459141) from August 2026, general availability in late September 2026. It is off by default. Third-party reporting mentions a BlockDetectedBots value for ExternalBotAccessMode, but as of 20 July 2026 the Learn page did not document it, so confirm the available values in your own tenant before relying on it. Blocking outright is simpler for admins but removes the organiser's judgement: with approval mode, a host can still admit a bot the company has approved.

Bot detection only helps if the bot actually waits in the lobby. Microsoft's own recommendation is to restrict who can admit people from the lobby to organisers and co-organisers. Otherwise any participant can let a detected bot — or anyone else — straight into the meeting, undoing the policy above.

The bots are driven by apps your users have consented to. In Microsoft Entra ID, require admin consent for apps so users cannot authorise a note taker on their own, and review the consents that already exist — sign-in logs will show which note-taking apps are active in your tenant. Note the limit: blocking an app in your tenant does not stop your users' bots joining meetings hosted by other organisations, where the host tenant's rules apply.

Google Meet now screens suspected bots into a separate risk queue that hosts see in the participant list; entries there are denied unless the host admits them. To reduce what reaches the queue at all, restrict anonymous joining in your Meet settings. Admins can also block specific note-taker apps' access to Google data under Security, then Access and data control, then API controls (verify the exact path in your admin console, as Google moves these settings). To control Google's own note taker, turn Gemini note-taking on or off per organisational unit: Apps, then Google Workspace, then Google Meet, then Gemini settings, then Gemini note-taking. Gemini note-taking is available in Business Standard and above.

Zoom has no single "block AI bots" switch — none was found in Zoom's own documentation. The working combination is: in Settings, then Security, enable "Block users in specific domains from joining meetings and webinars" and add the note-taker domains such as otter.ai and read.ai; require authenticated users to join meetings; and restrict Zoom Marketplace apps so only admin-approved apps can be installed. Be aware of the gap Zoom Community threads report: a bot joining through a user's authenticated account may still get past domain blocking, which is another reason to pair technical controls with policy.

People use note takers because they are genuinely useful, so a block without an alternative gets worked around. Approve one or more options: the native choices are Microsoft Copilot, Google Gemini and Zoom AI Companion, all governed by admin controls. A third-party option worth considering is a bot-free note taker such as Notta Desktop: because it records audio on the user's own computer rather than joining the meeting as a participant, it never appears in the lobby or participant list, works the same way across Teams, Zoom and Google Meet, and avoids the bot-governance problem entirely — you manage it through policy and device controls instead of meeting settings. For what to check before approving any tool, see our IT checklist for meeting recorders and our guide to transcription tools with strong access controls.

Write down which note takers are approved, which are banned, and who can approve exceptions. Tell organisers what to do when a bot appears anyway: remove it from the meeting, then report it so IT can follow up with the user or block the app. Publishing the policy is what makes the technical settings stick — without it, users simply pick tools the controls cannot see.