Microsoft to deploy AI-powered phishing detection in Teams

Microsoft is introducing a new AI-driven security feature for Microsoft Teams designed to detect phishing, social engineering, and impersonation attempts. Starting in early November 2026, the tool will use conversation context to identify suspicious behavior, even in messages that do not contain traditional indicators like malicious links or files. The rollout will begin with a targeted release phase through mid-November, with general worldwide availability expected between early and mid-December 2026.

The system will primarily focus on first-contact chats with external users. Microsoft has stated that the AI analyzes conversation patterns in real-time for security purposes only; message content will not be retained for model training, advertising, user monitoring, or other secondary uses. IT administrators will not need to manually enable this feature, as it is designed to deploy automatically to protect users from deceptive intent and identity theft.