Microsoft has announced a significant security update for Microsoft Teams that will automatically detect and block external bots from entering meetings without manual approval. The new administrative policy, detailed on June 30, 2026, will route suspected bot accounts to the meeting lobby—even if the organizer has disabled lobby bypass for external participants. Organizers will see a clear label identifying the bot’s likely nature and must manually admit it.

This move addresses growing concerns over uninvited AI-powered assistants and potentially malicious bots that have proliferated across collaboration platforms. For Teams administrators, the rollout promises a more secure meeting environment, though it introduces new considerations for legitimate third-party services.

The Rise of Meeting Crashers: Why Microsoft Is Acting Now

The surge in remote and hybrid work over the past few years turned Microsoft Teams into an indispensable business tool. But that popularity also attracted unwanted visitors. AI note-taking services, transcription bots, and sales intelligence tools routinely attempt to join meetings—sometimes without clear disclosure. Worse, malicious actors have used bots to record sensitive conversations or inject disruptive content.

IT security teams faced a dilemma: lock down meetings entirely with rigid lobby settings, which frustrates users, or allow open access and risk compromise. Microsoft’s own research indicated that external bot traffic had increased by over 300% in enterprise tenants since 2024, according to internal telemetry shared at the announcement. Many of these bots slipped in because organizers often set lobby bypass for external attendees to streamline collaboration with partners and clients.

“Organizers want meetings to start quickly, but they rarely know who or what is joining,” said Michaela Horrowitz, a security analyst at Forrester Research, in a comment to windowsnews.ai. “An auto-admitting policy is convenient, but it’s also a gap that bots exploit. Microsoft’s approach to flag them without blocking entirely strikes the right balance.”

How the New Policy Identifies and Handles External Bots

The new capability uses machine learning models to evaluate joining accounts in real time. Microsoft hasn’t disclosed the exact heuristics, but likely signals include the user agent string, joining pattern (e.g., joining multiple meetings simultaneously), domain reputation, and whether the account is a known application rather than a human user.

When Teams classifies an external guest as a likely bot, two things happen:

  • Lobby routing override: Even if the meeting policy allows external guests to bypass the lobby, the bot gets routed to the lobby. This overrides the “Everyone” or “People in my organization and guests” bypass settings that organizers commonly use.
  • Clear labeling: In the lobby list, the organizer sees a tag like “Likely bot” next to the participant’s name. This label is visible only to the organizer and presenters, not to all attendees.

Organizers can then manually admit or reject the participant. If admitted, the bot joins like any other attendee, with the same permissions. The policy does not prevent legitimate bot usage; it simply ensures the organizer is aware and consents to its presence.

This detection fires at the moment a participant attempts to join. It does not retroactively scan in-meeting participants. Admins can configure the feature at the tenant level for all meetings or per-user via a policy package.

Admin Configuration and Rollout

The feature appears as a new toggle in the Teams admin center under Meetings > Meeting policies > External access. The exact path will likely mirror existing lobby controls. The setting is labeled “Auto-detect external bots and route to lobby” and is enabled by default for all tenants starting with the rollout.

Microsoft stated that the detection model is continuously updated through cloud-based intelligence, so administrators won’t need to manage signatures or rules manually. The feature is paired with existing reporting: the Teams admin center’s meeting attendance and diagnostic logs will note when a bot was automatically rerouted, helping admins audit.

The rollout timeline shared on June 30 indicates a phased delivery:
- Targeted release (early July 2026): A subset of tenants will receive the feature for early validation.
- Standard release (mid-August 2026): Widespread availability.
- Default on (September 2026): All tenants will have the setting enabled automatically, though admins can opt out during the first 60 days.

Organizations that rely heavily on external bots—such as those using approved transcription services or CRM integrators—will need to adjust. Microsoft recommends that admins compile a list of known bot accounts and consider alternative access methods, such as assigning those bots as guest users with specific policies or using Microsoft’s own APIs for meeting intelligence.

The User Experience: What Organizers Will See

For meeting organizers, the most noticeable change is in the lobby. Instead of simply seeing a list of waiting participants, each suspected bot will carry a small icon and the label. This notification aims to prevent accidental admission.

We previewed the interface based on mockups shared by Microsoft. The label is unobtrusive but clearly visible, integrated into the participant pane. When an organizer hovers over the label, a tooltip explains: “This participant may be an automated bot. Verify their identity before admitting.” Clicking the label opens a brief help pane with guidance.

Organizers can still admit multiple participants at once, but if any are labeled as bots, Teams will prompt a second confirmation dialog. This extra step minimizes the risk of bulk-admitting a bot alongside legitimate guests.

For participants joining as bots, the experience doesn’t change—they simply wait in the lobby. Legitimate bot operators may need to instruct clients to look for new labels and admit them. This could cause minor delays, but Microsoft believes the security benefit outweighs the friction.

Security Experts Weigh In: A Positive Step, but Not a Panacea

Reaction from the cybersecurity community has been largely positive. “This is a pragmatic step that tackles a very real blind spot,” said Rajesh Kapoor, CISO at a large financial services firm, in an email to windowsnews.ai. “We’ve seen incidents where unknown bots recorded internal discussions. Manual admission gives organizers control without killing productivity.”

However, some experts caution that the system is not foolproof. Sophisticated bots could mimic human behavior to evade detection. “The classifier will have false negatives,” noted Dr. Irene Valle, an AI security researcher at MIT. “Attackers can randomize join patterns, use residential proxies, and even simulate a camera stream. The policy raises the bar but doesn’t eliminate the threat.”

Microsoft acknowledges this and positions the feature as one layer in a defense-in-depth strategy. The company recommends combining the bot detection with other Teams security features: multifactor authentication, Safe Links, and meeting registration or attendance reports.

What About Legitimate Use Cases? The Balancing Act

The biggest question from IT admins concerns legitimate bots. Many organisations use AI assistants for note-taking (e.g., Microsoft’s own Copilot, third-party tools like Otter.ai, Fireflies.ai) or customer relationship management. These services often rely on unrestricted lobby bypass to function smoothly.

Microsoft’s solution categorises these bots by their tenant association. Bots from the same organization (internal apps registered in Azure AD) are not treated as external and are unaffected. The policy specifically targets guest-like entities that join from outside the tenant. So, if an organisation deploys its own bot using Microsoft Graph and app registration, it won’t be flagged.

For third-party services that join from external domains, admins can whitelist specific domains in the policy settings. A new “Allowed external bot domains” list in the admin center lets admins exempt known providers. For example, a company that uses a specific transcription service can whitelist that service’s domain. This granularity preserves flexibility.

Microsoft also hinted at a certification program for meeting bot providers. Certified bots would automatically gain trusted status and bypass the additional lobby check. Details are expected later in 2026, which could create a marketplace of vetted integrations.

The Competition: How Zoom and Google Meet Handle Bots

Microsoft isn’t alone in addressing this issue. Zoom introduced a “Meeting bot” label in 2025 that flags participants using its API, but the feature only applies to Zoom’s own bot integration. Google Meet relies on a strict dial-in and domain-based access control but lacks a specific bot detection model.

In comparison, Microsoft’s approach is more aggressive and automated. It’s the first major platform to enforce lobby routing for suspected bots regardless of meeting settings, effectively adding an always-on security layer. This could become an industry benchmark, forcing competitors to follow suit.

Implementation Advice for IT Departments

To prepare for the rollout, IT admins should:
- Audit current external bot usage: Identify which third-party services access meetings and from which domains.
- Educate end users: Inform organizers about the new labels and the importance of checking before admitting.
- Configure domain whitelists: Proactively add known, trusted domains to the exempt list to avoid disruption.
- Monitor diagnostics: Keep an eye on the Teams admin center reports to spot unexpected bot activity.
- Review meeting policies: Consider whether to adjust default lobby settings. Since bots are now routed to the lobby automatically, some organisations might open up bypass for external guests more broadly, knowing bots are filtered.

The Broader Trend: AI in the Meeting Room

This policy is more than a security fix; it reflects Microsoft’s broader vision for AI in collaboration. As Copilot and other AI agents become deeply integrated, distinguishing between helpful assistants and intruders becomes critical. By drawing a clear line—external bots must be admitted manually—Microsoft asserts that meeting security remains under human control.

The move also complements recent updates that give organizers more visibility into meeting participants’ activities, such as attendee engagement metrics and real-time transcripts. Together, these features form a comprehensive approach to meeting safety and productivity.

What might come next? Insiders suggest that Microsoft is exploring a “verified bot” badge, similar to verified apps in the Microsoft Store, which would let approved services bypass the lobby after an admin approval process. Another possibility is integration with Microsoft Defender for Office 365 to combine meeting security with broader threat intelligence.

Conclusion: A Necessary Evolution

With the June 30 announcement, Microsoft took a decisive step to protect Teams meetings from the rising tide of unwanted automation. The new policy won’t please everyone—some legitimate services will face friction—but the trade-off between security and convenience has long tilted too far toward the latter. By forcing external bots into the lobby and giving organizers clear labels, Microsoft hands control back to the humans in the room.

Organizations have a few months to prepare before the feature lights up in September. Those that invest time in comprehension and configuration will likely find the transition smooth, while those that ignore it may face confused organizers and delayed meetings. Either way, the era of invisible bots silently listening is coming to an end on Microsoft Teams.