Microsoft 365 email security: what Defender covers and what to add

Yes. Every Microsoft 365 subscription with cloud mailboxes includes Microsoft’s built-in protection, still widely known as Exchange Online Protection (EOP). Microsoft documents anti-malware, anti-spam with bulk-mail protection, anti-spoofing with spoof intelligence, connection filtering, quarantine, zero-hour auto purge, and preset Standard and Strict policies. Microsoft publishes service levels for it: more than 99% spam effectiveness, a false-positive ratio below 1 in 250,000, and 99.999% monthly uptime.

Defender for Office 365 Plan 1 adds user and domain impersonation protection, mailbox intelligence (Microsoft’s contact graph), Safe Attachments, Safe Links and protection for Teams, SharePoint and OneDrive. Plan 2 adds Threat Explorer, Campaigns, advanced hunting, Automated Investigation and Response and Attack simulation training.

What changed in 2026

Microsoft began rolling Defender for Office 365 Plan 1 into Microsoft 365 E3/G3 and Office 365 E3/G3 on 17 June 2026. Microsoft’s licensing page says the features are complete by 1 August 2026; the Defender blog says “by Fall 2026”. The same update raised list prices from 1 July 2026 and added URL time-of-click protection to Office 365 E1 and Business Basic and Standard. If your last evaluation of Microsoft 365 email security predates June 2026, redo it: you may already own more than you think.

For bulk phishing, spam and known malware, Microsoft’s stack is strong, and Sentaro says so: “Microsoft Defender and Google’s built-in protection are a good baseline for bulk attacks, and Sentaro is designed to run with them rather than replace them.” The question is what lands after the filter has done its job.

Three kinds of attack are hard for any reputation- and payload-centered filter. First, business email compromise with no link or attachment: a credible instruction from the wrong person. Second, vendor email compromise, where the message comes from a real supplier’s real mailbox inside a real thread, so authentication passes. Third, consent phishing, where the user grants a malicious app access to their mailbox and no password is ever stolen. Defender’s impersonation protection helps with the first. The third is handled by a different Microsoft product, Defender for Cloud Apps, not by Defender for Office 365.

Sentaro’s Message vector reads intent (payment redirects, credential prompts, requests for secrecy and artificial deadlines), sender identity and infrastructure, and relationship and thread context, including suppliers that suddenly change banking details. Its App vector inventories every app, AI tool and OAuth grant in the tenant and revokes risky grants. Both are live today.

Do I need a third-party spam filter for Office 365?

Usually not for spam itself. Microsoft publishes a spam-effectiveness SLA above 99% for its built-in protection. Sentaro is not a spam filter and does not try to be one. If spam volume is your complaint, tune the preset Strict policy first. If the complaint is “the one email that got through was the one that mattered”, that is a detection problem a second, behavior-aware layer is built for.

Sentaro uses API-level integration via the Microsoft 365 admin SDKs. An admin connects the tenant, and protection starts immediately, with historical visibility within 15 minutes. There are no agents on endpoints, no MX-record changes and no rules to migrate, so mail keeps flowing through Microsoft exactly as before.

Because Sentaro sits inside the tenant rather than in front of it, it sees internal and historical mail, not only inbound. That matters after an account is compromised, when the phishing comes from a colleague. It also means Defender stays your first filter and Sentaro acts as the second opinion on what got through.

The exact Microsoft 365 permission scopes are not yet in Sentaro’s public documentation. The table lists what Sentaro does with its access, as published. Ask for the scope list and the DPA, and review them with whoever approves enterprise applications in your Entra ID tenant.

What do users see in Outlook?

Policy decides. For each user group, a detection maps to an action: block, quarantine or tag inline. Tagged messages are labeled in the mailbox, so the recipient sees the verdict where they read the mail rather than in a separate portal. Malicious messages are removed or quarantined automatically, and admins see the reasoning behind every verdict in the console.