Microsoft Teams will add AI checks for phishing and fraud in chats
Microsoft is adding an AI security feature to Teams that looks for phishing, scams and identity theft in conversation content. The rollout starts in early November 2026, with general availability expected in the first half of December.
Source: Viktor Eriksson
Microsoft will begin rolling out an AI-powered security feature in Teams in early November 2026 to help detect phishing, scams and identity theft in chats, according to company information cited by Computerworld. The tool is designed to inspect conversation content for suspicious behavior, with general availability expected in the first half of December 2026.
How the new protection works
Unlike older filters that focus mainly on suspicious links and files, Microsoft says the new feature will look at the messages themselves. The company said it is intended to spot behavior that may indicate fraud or impersonation, especially in conversations with new contacts, which Microsoft considers a higher-risk situation.
According to Microsoft, the feature may initially operate in the background while it goes through quality checks. After that stage, users may start seeing warnings when possible phishing is detected.
- Focuses on conversation content, not just links or attachments
- Targets interactions with new contacts first
- May run silently during early validation
- Can show warnings after validation is complete
Why Microsoft is adding it now
The move comes after Microsoft warned about attackers posing as help desk staff in Teams in an attempt to trick people into giving them remote access. The company’s new security measure appears aimed at reducing the chance that users will be fooled by convincing chat-based impersonation attempts.
The source says the feature is part of Microsoft’s broader effort to address phishing and fraud inside Teams, where attackers may rely on social engineering rather than malicious files or links. For ordinary users, that means some suspicious conversations could be flagged before they lead to a compromise.
What Microsoft says it will not do
Microsoft says the tool is meant only for security, not for surveillance or advertising. The company also said no messages will be saved or used to train AI models.
Another point for IT departments: Microsoft says administrators will not need to configure the feature. It will be turned on automatically when it becomes available.
Key dates
- Rollout begins: early November 2026
- General availability: first half of December 2026
Computerworld’s Viktor Eriksson reported the details, citing Microsoft and a Microsoft 365 Message Center post. The story was also referenced from Computer Sweden via Bleeping Computer.
Source: Viktor Eriksson
0 comments
No comments yet
Start the conversation.