In a Microsoft 365 Message Center post (mirror), the company said, “During initial quality validation, detection may run in the background without showing warnings to users. After validation, users may see a warning when potential phishing is detected.”
According to Microsoft, no messages will be saved or used to train AI models. The tool will also be used solely for security purposes — not, for example, for surveillance or advertising.
The rollout will begin in early November 2026 and is expected to be generally available during the first half of December. IT teams do not have to configure the feature; it will be automatically enabled.
This story originally appeared on Computerworld
