The operational distinction matters: existing tenants remain opt-in, while Microsoft is rolling out mandatory enablement for new tenants. Administrators should therefore treat this as both a security-hardening opportunity and a mail-flow planning exercise. Before closing the door, find out which legitimate applications are still walking through it.
What Reject Direct Send actually blocks
Direct Send lets on-premises devices, applications and third-party services send messages directly to Exchange Online-hosted mailboxes using an organization’s accepted domain, without authenticating as a mailbox user. Microsoft’s device-and-application guidance distinguishes this from authenticated SMTP submission and connector-based SMTP relay. Direct Send requires TCP port 25, does not require a licensed sending mailbox and cannot relay messages to recipients outside the Microsoft 365 organization.
That makes it useful for some legacy workflows, but also means legitimate business traffic can resemble unauthorized mail claiming the organization’s identity.
Microsoft Learn defines the rejection conditions precisely. With RejectDirectSend enabled, Exchange Online rejects incoming anonymous messages when:
- The message does not match an inbound connector configured to identify the sender by IP address or certificate.
- The domain in the SMTP
MAIL FROMaddress is an accepted domain in the receiving organization.
This is not a blanket ban on anonymous internet email. Nor does it inspect every address a user sees in Outlook. The distinction between the two sender identities is essential:
| Sender identity | Meaning | Evaluated by Reject Direct Send? |
|---|---|---|
| P1 envelope sender | The address supplied through SMTP MAIL FROM | Yes |
| P2 From header | The message’s From header | No |
Microsoft’s announcement explicitly limits the feature to P1 evaluation. Consequently, a message’s visible From address alone cannot establish whether this control should reject it. Other anti-spoofing protections remain necessary, including for impersonation involving domains the organization does not own.
Two further boundaries deserve attention. Subdomains are included when the accepted domain has Accept mail for all subdomains enabled, represented by MatchSubDomains = True. The control also works regardless of where the organization’s MX record points, so using a third-party mail gateway does not make this setting irrelevant.
What changed between preview and general availability
The release is more than a change of label. Microsoft highlights three improvements: the now-generally-available Change Optics Report, resolution of a known issue affecting Azure Communication Services Email, and clearer documentation covering permissions, subdomains, connector attribution and third-party MX configurations. The announcement does not explain the ACS issue’s underlying cause or provide a configuration-specific workaround.
The reporting improvement is particularly important. Enabling a rejection policy is straightforward; identifying everything it might interrupt is the harder work.
Microsoft separately announced general availability of the Change Optics Report on October 19, 2026. Its purpose is to show samples of messages potentially affected by Exchange Online service changes. At GA, its scenarios cover Direct Send traffic, labeled DRS, and external traffic using onmicrosoft.com addresses, labeled OMC. For this rollout, administrators should concentrate on DRS.
The report is available under:
Exchange admin center → Reports → Mail flow → Change Optics Report
Its Summary page shows message-volume trends, while Details provides example messages for investigation. Microsoft warns that these are samples, not an exhaustive message trace. There is also an export limitation: exporting captures data loaded into the table, rather than necessarily everything in the selected time range. Microsoft has described plans to improve that behavior without providing an ETA.
The practical conclusion is straightforward: use the report to discover sending patterns, not as proof that every dependency has been accounted for.
Existing tenants and new tenants have different rules
For existing tenants, Microsoft says Reject Direct Send remains off by default and administrators choose whether to enable it. That is the default policy, not a statement that an organization which enabled the preview has had its setting reset. Read the actual tenant configuration before planning a change.
For new Exchange Online tenants, Microsoft says it is rolling out a different model: Reject Direct Send enabled by default, without an option to disable it. The November 16 announcement describes that deployment as underway, but provides neither a completion date nor a tenant-by-tenant schedule. It should not be read as confirmation that every newly created tenant already has the restriction.
Availability also needs careful separation:
- Reject Direct Send: Generally available in the worldwide environment; Microsoft explicitly excludes GCC High and DoD.
- Change Optics Report: Generally available in the worldwide environment; its separate announcement excludes GCC, GCC High, DoD and other sovereign cloud environments.
Do not assume the reporting tool and enforcement setting have identical cloud availability.
This is an Exchange Online service setting, not a Windows cumulative update or an on-premises Exchange patch. Microsoft Learn identifies RejectDirectSend as a cloud-only parameter. There is no client-side Windows build requirement attached to that parameter.
A practical deployment sequence
The safest implementation sequence is discovery, sender remediation, attribution testing and then enforcement. The following workflow combines Microsoft’s deployment instructions with its PowerShell and message-trace documentation.
1. Prepare the administrative session
Use an authenticated Exchange Online PowerShell session with the necessary permissions. Microsoft specifies the Organization Configuration role for changing Reject Direct Send. The message-trace commands also require appropriate permissions; having a working PowerShell connection is not, by itself, evidence that every command is authorized.
There is a separate tooling prerequisite worth checking: Microsoft Learn requires Exchange Online PowerShell V3 module version 3.7.0 or later for Get-MessageTraceV2. This is a requirement for the verification workflow, not a newly announced minimum version for the rejection setting itself.
Read the current configuration before making changes:
Get-OrganizationConfig | Format-List RejectDirectSend
Record the returned value in the change request. This avoids treating a tenant already protected during preview as a fresh deployment.
2. Inventory legitimate Direct Send sources
Review the DRS scenario in Change Optics and investigate the devices, applications and services represented there. Microsoft also recommends examining your domains’ SPF records as another starting point for identifying authorized senders.
For the change plan, build a working inventory containing each sender’s owner, business purpose, sending domain and available certificate or source-IP information. This is an administrative recommendation: those details help translate a message sample into a dependency somebody can validate.
Do not confuse an SPF entry with connector attribution. Microsoft’s parameter reference bases the Direct Send decision on the envelope domain and whether an IP- or certificate-matching connector identifies the incoming message. Merely listing a sender in SPF does not satisfy that connector condition.
3. Establish connector attribution
Microsoft directs administrators to create a partner inbound connector for each legitimate sender that would otherwise be rejected. The connector should match the sender’s TLS certificate, the recommended approach, or its sending IP addresses.
The objective is not simply to create a connector object. It is to establish that messages arriving from the actual production sending path match it.
Microsoft’s broader SMTP relay guidance reinforces why sender identity matters: certificate-based or public-IP-based identification distinguishes connector-authenticated delivery from anonymous Direct Send. For an organization’s own IP-based relay configuration, Microsoft requires a static public IP that is not shared with another organization. Those relay requirements should not be confused with a universal configuration recipe for every third-party partner.
4. Verify with real test messages
Send a test message from each legitimate source and inspect its trace details. Microsoft supplies this example:
Get-MessageTraceV2 -MessageId "<message ID>" |
Get-MessageTraceDetailV2 |
Format-List
The intended check is whether the message was attributed to the expected connector before enforcement is enabled.
Replace the placeholder with the message’s full Message-ID. Microsoft’s cmdlet documentation says to retain the complete string, including angle brackets if they are part of the identifier, and enclose it in quotation marks. MessageId is not the same identifier as the service-generated MessageTraceId.
If investigating older traffic, account for trace limits rather than assuming an empty result proves there was no message. Microsoft documents 90-day retention, a maximum 10-day interval per query and a default 48-hour result window when the cmdlet is run without parameters.
5. Enable enforcement and validate the outcome
Once legitimate sources are accounted for, enable the setting:
Set-OrganizationConfig -RejectDirectSend $true
Microsoft says propagation across the service takes up to 30 minutes. Then check the configuration again:
Get-OrganizationConfig | Format-List RejectDirectSend
The setting should report True.
Configuration success and operational success are different checks. As a deployment recommendation, repeat the legitimate-sender tests after propagation and confirm both expected delivery and connector attribution. A controlled test that meets the rejection conditions should receive Microsoft’s documented response:
550 5.7.68 TenantInboundAttribution; Direct Send not allowed for this organization from unauthorized sources
That specific response identifies this enforcement path; an unrelated delivery failure should not automatically be blamed on Reject Direct Send.
For an existing tenant where the setting remains configurable, Microsoft Learn documents $false as disabling the block:
Set-OrganizationConfig -RejectDirectSend $false
Treat rollback as a documented contingency, not the preferred permanent fix for an unidentified sender. It is not an escape route to promise for new tenants covered by the mandatory-default rollout.
Two caveats that can derail an otherwise clean rollout
Forwarding can bring your own sender address back
Microsoft identifies a known issue involving third-party forwarding without Sender Rewriting Scheme, or SRS.
The sequence is simple: someone in your organization sends a message externally; the recipient’s provider forwards it back into your organization without rewriting the original envelope sender. The returning message consequently still claims your domain. Microsoft says these messages already fail SPF, and Reject Direct Send will reject them when no partner connector attributes the returning traffic.
The actionable lesson is to include return-forwarding workflows in testing. Testing only newly generated messages from applications will not exercise this separate path.
Some vendors cannot supply a suitable connector identity
The fetched GA announcement also addresses vendors without static IP addresses or a dedicated certificate. Microsoft suggests considering a mail-flow rule with an exception—for example, involving a unique X-header—instead of the organization-wide setting.
That wording matters. It is an alternative policy approach, not documentation that a header exception overrides Reject Direct Send after enforcement is enabled. Administrators should evaluate such a design separately rather than assuming they can combine the two into an automatic bypass.
The bottom line
Reject Direct Send provides a narrowly defined control: reject unauthenticated incoming messages using the organization’s accepted envelope-sender domains unless the source is appropriately attributed. It does not replace broader anti-spoofing controls, and legitimate legacy senders can fall within its scope.
For existing tenants, the sensible next step is a dependency review followed by tested enablement. For new deployments, plan around Microsoft’s move toward mandatory rejection rather than assuming anonymous Direct Send will remain available.
The PowerShell change is one line. The valuable work is establishing which senders deserve to keep delivering—and proving that Exchange Online recognizes them before enforcing the boundary.
References
- Reject Direct Send now Generally Available in Exchange Online Microsoft Exchange Team Blog · Mon, 16 Nov 2026 16:18:57 GMT
- Reject Direct Send now Generally Available in Exchange Online | Microsoft Community Hub techcommunity.microsoft.com
- How to set up a multifunction device or application to send email using Microsoft 365 or Office 365 | Microsoft Learn learn.microsoft.com