Microsoft Teams will use AI to spot scam messages from strangers

Microsoft Teams is getting a built-in scam checker. Starting in early November 2026, AI will look at chat messages for signs of phishing, which means tricking someone into handing over information or money. It will focus on first-contact chats, where someone outside your company messages you for the first time.

What changed

Microsoft calls the feature AI-powered, context-aware phishing detection. It reads the flow of a conversation to spot phishing, social engineering, impersonation and deceptive intent. Social engineering means manipulating a person rather than hacking a computer.

The key point is that it does not need a bad link or file to raise a flag. Microsoft says it can catch suspicious behavior that lacks the usual warning signs.

Privacy and setup

Microsoft says IT admins do not need to turn the feature on. For most users it will watch external first-contact chats.

On privacy, Microsoft is direct about limits. It says the tool is used only for security protection, and not for productivity, summarization, advertising or user monitoring. It also says customer message content is not kept by the detection system for model training or other secondary uses.

When it arrives

• Targeted release starts in early November 2026 and is expected to end by mid-November.
• General availability worldwide is likely between early and mid-December.

Part of a bigger push

This is one of several recent Teams security changes aimed at outside attackers. Microsoft has said admins can block external users through the Defender portal, and a meeting policy lets admins automatically block identified external bots.

It has also said Teams will blur QR codes sent by external senders and let users report suspicious guest invitations. In early October, it announced plans to support third-party deepfake detection and impersonation protection in Teams meetings.

What to do

If your business uses Teams, the checks should turn on by themselves. They are an extra layer, not a guarantee. Staff should still treat unexpected chats from outsiders with caution, especially ones that ask for information, money or quick action. The material does not say how accurate the detection is or what warnings users will see.