Microsoft Teams to add AI phishing detection and open live meetings to third-party deepfake detectors in November

MSFT is adding two AI-era security features to Teams. The first is AI-powered phishing detection in Teams messaging that uses conversation context to spot phishing and social engineering, focused mainly on higher-risk chats where someone outside the organization makes first contact; users may see warnings when suspicious activity is detected. Microsoft's 365 roadmap lists it as in development with general availability targeted for November 2026, across desktop, web and mobile.

The second lets organizations plug certified third-party deepfake detectors into live Teams meetings. According to Microsoft's message center notice, Targeted Release begins in mid-November and should finish by late November, while broad availability has no date yet. An administrator picks a provider, deploys its app in the Teams admin center and grants permissions; the provider analyzes meeting audio and video and sends a detection signal back, which Teams shows to participants. The detection itself is done by the third party, and Microsoft has not named any certified vendors.

The moves respond to a real shift in enterprise risk: voice cloning and face swapping have made impersonation on video calls a practical attack. By opening a certification path rather than building only its own detector, Microsoft creates a new distribution channel for security vendors on one of the largest collaboration platforms, although Windows Latest notes a configuration hint that a first-party Microsoft detector may follow.

What to watch: which vendors get certified, what licence tier the features require, and how accurate the detectors prove in practice. Microsoft has not published false-positive rates, data-retention terms or whether participants are told a detection bot is present.