AegisAI Perspectives · 143
Microsoft 365 filters mail before delivery, and AegisAI reads it through the API after delivery, with removal access that your admin grants and revokes.
See it on your own Google Workspace or Microsoft 365 tenant. No MX change.
Or get the Q3 2026 State of the AI Attack report (free, instant download)
Not ready for a demo? Check your domain.

Microsoft 365 already protects every cloud mailbox against spam, malware, phishing and other email threats1. AegisAI investigates every email for what it is trying to do, and removes the ones that fail from every mailbox they reached18. AegisAI reports on its inbound email security page that 3 in 10 confirmed phishing emails cleared native Gmail and Microsoft filters in Q3 202624.
If you run IT for a company of 20 to 500 people on Microsoft 365, with a small IT or security team and no SOC, this page shows what your licence already covers and where it stops. It also lists the mailbox access the added layer needs, and how you limit it. Which product to pick is in email security for Microsoft 365: settings first, then what to add. How the API connection is set up is in what API-based email security is and how it integrates with Microsoft 365.
Key takeaways
- Built-in protection against email threats covers all cloud mailboxes.
- Defender for Office 365 Plan 1 comes with Microsoft 365 E3 and Business Premium; Plan 2 comes with E523.
- Standard and Strict preset policies are likely off on your first visit to the Preset security policies page6.
- Zero-hour auto purge searches only the last 48 hours of delivered email7.
- Application Mail.ReadWrite reads and deletes email in all mailboxes but does not include permission to send mail14.
| Question | Answer | Source |
|---|---|---|
| What is API-based email security and how does it integrate with Microsoft 365? | A layer that reads mail through the Microsoft 365 API after delivery, with no MX change and no gateway19. It reaches mailboxes through Microsoft Graph application permissions such as Mail.ReadWrite, which an admin can limit with RBAC for Applications (role based access control for apps in Exchange Online)13. | Inbound email security page; Microsoft Learn |
| What does Microsoft 365 protect by default? | Spam, malware, phishing and other email threats in all cloud mailboxes. With at least one Defender for Office 365 licence, the Built-in protection preset adds Safe Links and Safe Attachments910. | Microsoft Learn |
| Which licence adds what? | Plan 1 (E3 and Business Premium) adds impersonation protection plus Safe Links and Safe Attachments4. Plan 2 (E5) adds automated investigation and response11. | Microsoft Learn |
| What does the added layer do after delivery? | It investigates every email for what it is trying to do and removes the ones that fail from every mailbox they reached. The median verdict takes 1.9 seconds, per the inbound email security page20. | Inbound email security page |
| Who controls its access to mailboxes? | Your admin. Removal needs write access your admin grants and can revoke25. Revoke it on the Admin consent tab in Entra ID16. | Vendor API access page; Microsoft Learn |
Short on time? Watch the explainer, or book a demo.
It covers every cloud mailbox with built-in protection against email threats. Defender for Office 365 Plan 1 is included in Microsoft 365 E3 and Business Premium. Plan 2 is included in E5. Plan 1 adds user and domain impersonation protection to anti-phishing policies. It also adds mailbox intelligence and phishing email thresholds.
There is no default Safe Links policy and no default Safe Attachments policy. The Built-in protection preset fills that gap for all recipients when the tenant has at least one Defender for Office 365 licence. It covers only users who are not in the Standard or Strict presets or in custom Safe Links or Safe Attachments policies5. Standard and Strict protection are likely turned off on your first visit to the Preset security policies page.
The gaps begin at three edges. The first is time. Zero-hour auto purge (ZAP) acts on delivered mail, but its search is limited to the last 48 hours of delivered email. ZAP has no special licensing requirements for malware, spam or phishing8, so every Exchange Online mailbox gets the same 48 hour window.
The second is people. Automated investigation and response (AIR) needs Plan 2. When AIR identifies threats, it queues remediation actions for SecOps personnel to approve12. A team with no SOC has to staff that queue.
The third is mail with nothing to scan. Safe Links and Safe Attachments act on links and files. A payment request with no link and no file gives them nothing to open. Business email compromise losses reported to the FBI's Internet Crime Complaint Center were $3,046,598,558 in 2025, per the 2025 IC3 Annual Report17.
Open the Preset security policies page in the Microsoft Defender portal and write down which users each preset covers.
An API layer fits on top of Microsoft 365 without touching mail routing. It sits out of band: it reads mail through the Microsoft 365 API after delivery, with no MX change and no gateway. An inline gateway sits in the mail path instead, so every message passes through it on the way in.
The two architectures act at different times. The API layer acts after delivery and removes what fails from every mailbox it reached. Keep every Microsoft 365 setting on whatever you add.
You control the permission, the mailboxes it reaches and when it ends. An API layer reaches mail through Microsoft Graph, so its reach is the permission you grant. Consent roles and setup steps are in what API-based email security is and how it integrates with Microsoft 365. These steps set the limits.
Read the permission before you consent. Application Mail.ReadWrite allows the app to create, read, update and delete email in all mailboxes without a signed-in user, and does not include permission to send mail.
Grant read access first, and hold write access until you have reviewed the verdicts.
Scope the mailboxes. RBAC for Applications in Exchange Online lets admins grant permissions to an application that accesses Exchange Online data on its own.
Remove the wide grant. Microsoft states you need to remove the organization-wide unscoped permissions assigned in Microsoft Entra ID, or the scope does not hold15.
Review the grant in the Microsoft Entra admin center: Entra ID, Enterprise apps, All applications, the app, Permissions, Admin consent tab.
Revoke it from the same tab when a trial ends or the app is retired.
Do steps 1 to 4 before you grant write access to any app.
It does not change the facts above, and this page makes no claim about any law, standard or framework. A compliance review of an API layer starts from three facts you can check. The permission it holds, and what that permission cannot do, such as sending mail. The mailboxes it reaches, set with RBAC for Applications. The reason behind each verdict, given in plain language21.
Take those three facts to your own compliance owner before you grant write access.
It sits after them: each layer acts at a different point and comes with a different licence, so check them one row at a time.
| Layer | Who gets it | Default | What it covers | Source |
|---|---|---|---|---|
| Exchange Online Protection | All cloud mailboxes | Built in | Spam, malware, phishing and other email threats | |
| Safe Links | Defender for Office 365 Plan 1 or Plan 2 | No default policy; Built-in protection preset | Links in email and Teams, and in supported Office files | |
| Safe Attachments | Defender for Office 365 Plan 1 or Plan 2 | No default policy; Built-in protection preset | Attachments for recipients outside Standard, Strict or custom policies | |
| Impersonation protection | Defender for Office 365 Plan 1 or Plan 2 | Set in anti-phishing policies | User and domain impersonation; mailbox intelligence; phishing thresholds | |
| Standard and Strict presets | The users you assign | Likely off on first visit | Every policy in the preset for those users | |
| Zero-hour auto purge | All Exchange Online mailboxes, no special licence | Searches the last 48 hours | Delivered malware, spam or phishing | |
| Automated investigation and response | Defender for Office 365 Plan 2 | Queues actions for SecOps approval | Threats found in an investigation | |
| The added API layer | Connected by API, no MX change | Read access first; removal needs write access the admin grants | Every email, after delivery | 23 |
Keep the Microsoft rows on and add the last row on top of them.
It reads mail through the Microsoft 365 or Google Workspace API after delivery. It catches fake executives, vendors and payment requests, even with no link or file to scan22. Every verdict comes with its reason in plain language, so your team can check it and share it. The inbound email security page states that the median verdict takes 1.9 seconds, and a convicted message is removed from every mailbox it reached within seconds.
Track three numbers, each from a place you can open.
To get the third number on your own mail, ask for a test on your own tenant. Put the three figures side by side on day 30 and decide on them.
It is a layer that reads mail through the Microsoft 365 API after delivery, with no MX change and no gateway. Access comes from Microsoft Graph application permissions an admin grants. AegisAI reads mail through the Microsoft 365 or Google Workspace API after delivery, and removal needs write access your admin grants and can revoke.
Yes, when the tenant has at least one Defender for Office 365 licence. There is no default Safe Links policy, but the Built-in protection preset covers email and Teams, and supported Office files, for all recipients.
No. Microsoft lists no special licensing requirements for ZAP against malware, spam or phishing. Its search covers only the last 48 hours of delivered email.
No. Application Mail.ReadWrite allows the app to create, read, update and delete email in all mailboxes, and does not include permission to send mail.
Because the Entra ID grant still applies. Remove the organization-wide unscoped permissions assigned in Microsoft Entra ID, then check the app's Admin consent tab.