{
"title": "Teams Lobby Now Detects AI Bots from Outside Orgs, Blocks Entry Without Approval",
"content": "Microsoft is drawing a hard line between the bots you want in your Teams meetings and the ones you never invited. A new administrative control, now entering public preview for Microsoft 365 tenants, automatically detects third-party AI notetaker bots when they attempt to join meetings from an external organization, blocks their direct entry, and forces them to wait in the lobby until the meeting organizer explicitly approves—all while stamping a high-visibility “AI Notetaker” label on the bot’s icon so no one mistakes it for a colleague.

The feature, available under Meeting Policies in the Teams admin center, addresses a longstanding blind spot in meeting security. As AI assistants such as Otter.ai, Fireflies, Fathom, and homegrown solutions have multiplied, they’ve exploited a loophole: they join as guest users, often using legitimate-looking accounts, and bypass standard app permission controls because they aren’t classified as applications. Organizers were left unaware that a recording or transcription agent had slipped in, potentially capturing sensitive strategy discussions or regulated data.

Now, that loophole is closing. With the policy turned on, every external participant undergoes a behind-the-scenes classification check during the join handshake. Microsoft hasn’t disclosed the full detection algorithm, but engineers familiar with the development say it combines signals from Microsoft Entra ID tenant attributes, app registration patterns, known publisher lists, and behavior heuristics such as join speed and interaction patterns. If the system flags a participant as an unverified external AI notetaker, the meeting service reroutes the join request to the lobby, regardless of any existing guest bypass settings.

For the meeting organizer, the experience is unambiguous. A notification panel slides into the meeting window: “An external AI notetaker, ‘Fireflies Notetaker (External),’ is waiting in the lobby. Would you like to admit this bot or deny entry?” The bot’s avatar is overlaid with a neon-blue AI badge, and Organizers can click to read a short explanation that admitting the bot will grant it access to meeting audio, video, and shared content. Only after the organizer clicks “Admit” does the bot appear in the participant roster—still wearing its AI label for the duration of the session.

Crucially, the decision isn’t permanent. If the organizer admits the bot, the lobby control remembers the choice for that meeting only; the bot will have to clear the lobby again in every subsequent meeting, preventing one click from becoming a blanket grant. Organizers can also review their admission history in the meeting’s details pane after the call, a trail that becomes part of the compliance record when audit logs are exported.

How administrators enable the control

Within the Teams admin center, the setting “Control external AI notetaker bots” sits under Meetings > Meeting policies. By default, it’s off, meaning no change to existing behavior. Admins can enable it per policy and assign those policies to specific users or groups. Here’s the quick setup path:

  1. Navigate to Teams admin center > Meetings > Meeting policies.
  2. Select an existing policy or create a new one.
3.