Microsoft Teams is preparing to give meeting organizers another way to spot AI fakes before a convincing face or voice turns a routine work call into a security incident.
The platform will support certified third-party tools that detect synthetic or manipulated audio and video during Teams meetings. The feature is listed as in development, with a worldwide rollout planned for next month.
The update will let detection providers analyze meeting media for signs of manipulation and send alerts to Teams. Microsoft's platform will then surface those signals through in-meeting experiences and controls. However, Microsoft is not building the detection technology itself. Instead, it will provide the integration that allows supported providers to deliver their findings within Teams.
The initial roadmap listing covers the desktop, Mac and web versions of Teams, with targeted release and general availability phases. Microsoft has not named the supported providers or explained whether organizations will need additional licenses to use their services.
A separate layer for impersonation warnings
Microsoft is also developing a separate impersonation protection feature for Teams meetings. It will warn users when Teams identifies a potentially deceptive meeting organizer or participant, displaying risk indicators to help people assess suspicious identities.
The two features address related but distinct problems. Deepfake detection focuses on manipulated audio and video, while impersonation protection focuses on suspicious identities. Together, they could help organizations identify warning signs during meetings that might otherwise appear legitimate.
The additions follow other Teams security measures, including blurring QR codes sent by external users, blocking identified external bots through meeting policies, and plans to let users report suspicious guest invitations directly from Teams.
Why businesses need more than a familiar face
Video meetings have become a potential entry point for social engineering, where attackers exploit trust rather than relying solely on software vulnerabilities.
A convincing face or voice can make fraudulent requests appear credible, particularly when employees believe they are speaking with a colleague, executive or IT support worker.
Deepfake detection could provide another layer of protection, but its effectiveness will depend on the technologies organizations deploy. Detection tools can produce false positives or fail to identify sophisticated manipulation, making their results useful warning signals rather than definitive proof of someone's identity.
Microsoft has not yet published details about detection accuracy, supported vendors or how the integration will handle meeting data.
Those unanswered questions will matter to security teams deciding whether the integration is appropriate for sensitive business meetings.
What Teams users should expect
For employees, the changes could provide another warning when a meeting participant's audio, video or identity appears suspicious. Administrators, meanwhile, will need to assess provider availability, licensing requirements and data-handling practices before deciding whether to deploy a detection service.
The November rollout date does not mean every organization will receive the features immediately, and the published roadmap does not specify an exact launch day or explain the full setup process.
Even with detection enabled, employees should independently verify unusual requests involving payments, credentials, confidential files or software installations. A warning may expose a suspicious interaction, but the absence of one does not establish that a person is authentic.
Other news: Samsung Galaxy S26 Hacked Again as Pwn2Own Researchers Find More Flaws. Security researchers successfully compromised the Galaxy S26 three more times during the second day of Pwn2Own Ireland, exposing additional vulnerabilities.