Microsoft Teams Adds Policy to Auto-Block External AI Meeting Bots
Microsoft Teams now lets administrators automatically block identified external meeting bots — AI note-takers and recorders invited by outside participants — from joining meetings, via a new meeting policy option tracked under Message Center post MC1459141 and Roadmap ID 566201. General availability began in early October 2026 and is expected to complete by the end of November.
The control lands in the existing “Manage bots” section of Teams meeting policies, is off by default, and does nothing until an administrator turns it on. Until now, Teams could only detect external bots, park them in the lobby and require the organiser to admit them. The new option removes the human decision: detected bots are simply denied entry.
The Tech TL;DR:
- A new Teams meeting policy option blocks identified external AI meeting bots outright, off by default, assignable per user or group.
- PowerShell administrators can set it via the
ExternalBotAccessModeparameter with the valueBlockDetectedBots. - Detection is imperfect — unidentified bots can still join, and humans may occasionally be misclassified — so Microsoft recommends piloting before tenant-wide rollout.
Why external bots became a compliance problem
External AI note-takers increasingly turn up in Teams meetings, invited by someone outside the organisation, and the recording or transcript often ends up in third-party systems with nobody checking where.
According to Microsoft’s official documentation on managing external bots, these tools can record or transcribe meetings without participant awareness and store meeting data outside the organisation’s compliance boundaries — a data-leakage and governance risk that sits outside most tenants’ SOC 2 and data-residency controls.
Teams’ detection mechanism combines infrastructural and behavioural signals collected during the join process, per the Microsoft Learn documentation. Detected bots are placed in the lobby regardless of the meeting’s lobby configuration, and organisers are warned about the risk before admitting them. Microsoft also advises restricting lobby admission to organisers and co-organisers only, so a presenter cannot wave a bot through unscrutinised.
Rollout timeline: GA completes by end of November
Targeted Release started in mid-September 2026 and was due to complete by late September. General Availability began in early October, and Microsoft updated the timeline on November 3: completion is now expected by the end of November 2026. If the toggle isn’t visible in a tenant yet, the advice is to wait rather than hunt for workarounds — the option may simply not have arrived.
The exact wording of the new toggle has not been confirmed across tenants, so administrators should look for the option to automatically block identified external bots inside the Manage bots section rather than expecting a specific label.
How to enable auto-block, step by step
The admin-center path: sign in to the Teams admin center, go to Settings & policies, then Meeting policies. For a pilot, create a new policy such as “Pilot – Block meeting bots” via Add; to apply tenant-wide, edit Global (Org-wide default). Scroll to Manage bots, turn on the automatic-block option, and save. Policy changes take a few hours to propagate — they aren’t instant.
Testing matters. Schedule a meeting as a pilot user and let an external bot attempt to join. Under the new setting it should be blocked outright, not left sitting in the lobby awaiting approval.
The PowerShell route: ExternalBotAccessMode
For scripted deployments, the Set-CsTeamsMeetingPolicy documentation lists an ExternalBotAccessMode parameter with three values, and the three map cleanly to escalating security postures:
| Admin Center option | PowerShell value | Behavior |
|---|---|---|
| Do not detect bots | AllowBots |
Bots aren’t detected; they appear as any other external participant. |
| Require approval when detected | RequireApprovalWhenDetected (default) |
Bots are marked in the lobby until explicitly admitted. |
| Block identified external bots | BlockDetectedBots |
Detected bots are prevented from joining meetings governed by the policy. |
A pilot policy via the Microsoft Teams PowerShell module:
Connect-MicrosoftTeams
New-CsTeamsMeetingPolicy -Identity "Pilot-BlockBots"
Set-CsTeamsMeetingPolicy -Identity "Pilot-BlockBots" -ExternalBotAccessMode BlockDetectedBots
Grant-CsTeamsMeetingPolicy -Identity "[email protected]" -PolicyName "Pilot-BlockBots"
Get-CsTeamsMeetingPolicy -Identity "Pilot-BlockBots" | Select-Object Identity, ExternalBotAccessMode
Parameter availability can lag the rollout, so cmdlets should be tested in the target tenant first.
Known limitations: detection isn’t perfect
The Microsoft Learn documentation states plainly that some external bots may not be detected, meaning an unidentified bot can still get in, and occasionally a human participant may be misclassified. The default RequireApprovalWhenDetected setting remains Microsoft’s recommended balanced posture, with admins advised to educate organisers, monitor audit logs for unusual participant activity, and avoid disabling detection outright.
Pilot first, then widen
Microsoft’s own recommendations: review which meeting bots and automated meeting applications the organisation actually uses, evaluate with a pilot group before broader deployment, review existing meeting policies to decide which users or groups get the setting, update internal documentation and help-desk guidance, and communicate the change to organisers who rely on automated note-takers.
The practical risk is breaking workflows people depend on. Blocking unapproved bots works best when paired with a clear position on which AI meeting tools are approved — the people being blocked need to know what they should use instead.
Enterprises without in-house Teams expertise may want a managed IT service provider or Microsoft 365 consultant to run the pilot and policy mapping; the work is straightforward PowerShell, but the blast radius of misapplied meeting policies is real.
Where Teams bot governance goes from here
The direction of travel is clear: Microsoft moved from visibility (mark bots in the lobby) to enforcement (block them) within a single release cycle, and detection coverage is described as continuously improving. What remains open is the detection gap itself — until Microsoft closes it, an unrecognised bot still walks past the policy. Administrators watching the Message Center post MC1459141 will see the GA rollout complete by the end of November 2026.
Microsoft was founded on April 4, 1975, by Bill Gates and Paul Allen to sell a BASIC interpreter for the Altair 8800, and rose to dominate the PC operating system market with MS-DOS in the mid-1980s before Windows.
Headquartered in Redmond, Washington, it is the largest software company by revenue, and since Satya Nadella became CEO in 2014 it has shifted towards cloud computing and artificial intelligence — the same AI focus now visible in meeting bots. In April 2019 it became the third publicly traded U.S. company valued at over $1 trillion.
Always consult with certified IT and cybersecurity professionals before altering enterprise networks or handling sensitive data.