Microsoft is preparing to introduce an artificial intelligence-powered security feature in Teams to detect phishing attempts, impersonation and scams by analyzing conversations, including messages that contain no malicious links or attachments.
Microsoft announced the update in a Message Center notice published on Oct. 21. The initial rollout is expected to begin in early November, with the targeted release scheduled to run through mid-November. Worldwide general availability is expected between early and mid-December. The feature will be available on desktop, Mac, web and mobile devices.
Traditional phishing protection often focuses on dangerous links, malicious attachments and suspicious files. However, scammers increasingly rely on social engineering, manipulating people into sharing sensitive information or granting access to their devices without sending harmful files.
Microsoft’s new feature will analyze conversations to identify signs of phishing, deceptive intent, impersonation and other suspicious behavior. This includes attacks in which criminals pose as technical support staff, colleagues or other trusted contacts to gain access to company systems.
READ: Microsoft offers buyouts to 60,000 workers as it shifts spending toward AI (October 26, 2026)
In an example shared by Microsoft, an external account using the name “MOD Administrator” discusses permission problems before asking, “Can you share your credential first?” Teams displays a warning that the message could be a phishing attempt. The example illustrates how an apparently legitimate conversation can be used to steal sensitive information.
The company said the new protection will initially focus on conversations with unfamiliar external contacts, including individual and group chats. Microsoft has also indicated that Teams mobile users in China may receive protection for certain internal conversations.
The feature will initially operate in the background while Microsoft validates its performance. Visible warnings may be introduced after that validation, meaning users should not expect every suspicious message to trigger an immediate alert during the rollout. Microsoft said administrators will not need to make additional configuration changes to enable the protection.
READ: Elon Musk welcomes Grok integration into Microsoft Copilot (September 13, 2026)
The company said the capability is intended solely for security purposes and will not be used for advertising, employee monitoring, productivity analysis or conversation summaries. Addressing privacy concerns, Microsoft said message content will not be retained by the detection system for AI model training or other secondary purposes.
Despite the additional protection, Microsoft advises organizations to continue educating employees about suspicious messages and unfamiliar contacts. Users who receive warnings should verify the sender’s identity through a trusted communication channel. They should avoid sharing passwords, verification codes, financial details or other sensitive information in unexpected support conversations.
The new feature adds another layer of protection against increasingly convincing scams, which generative AI can make easier to create. However, the same technology can also be exploited by scammers to craft persuasive phishing messages, making vigilance and verification essential.