Forensic Triage: Why a Phish Bypassed Your p=reject DMARC Policy
A rigid DMARC reject policy means nothing if an ancient allow-list in your mail flow rules explicitly overrides the verdict.

The alert drops in your queue indicating a high-confidence phishing campaign targeting executive assistants with a credential harvesting link. You pull the raw message headers, check the sender domain in your DNS lookup tool, and see the holy grail of email security: a DMARC record explicitly set to v=DMARC1; p=reject; sp=reject. The sender domain owner explicitly told the world to drop any messages failing authentication.
Yet the payload is sitting unquarantined in the target inbox. This is the exact moment junior analysts blame the protocol and senior analysts go hunting for the misconfiguration. DMARC is not a firewall. It is a declarative policy framework defined in RFC 7489 that relies entirely on the receiving Mail Transfer Agent to honor the requested enforcement action.
When a malicious email with a dmarc failed but delivered status hits your SIEM, the protocol itself is rarely broken. The failure lies in your defensive architecture. Somewhere in the stack of Secure Email Gateways, cloud provider routing rules, or legacy mail flow configurations, a local override told the receiving server to ignore the DMARC verdict and prioritize implicit trust over cryptographic proof.
Interrogating the Authentication-Results Header
The first step in any email triage is isolating the exact handoff where the DMARC evaluation occurred. This requires finding the authoritative Authentication-Results header stamped by your final inbound hop. Attackers frequently forge these headers to confuse parsers, so you must start from the top of the header stack and trace the Received lines backward until you hit your own infrastructure boundary.
Parsing the Disposition
You are looking for the specific key-value pairs dictated by RFC 8601. A true policy bypass scenario usually presents a stark contradiction right in the header string, where the mathematical failure is acknowledged but the enforcement is skipped.
Authentication-Results: mx.contoso.com; dkim=fail (body hash did not verify); spf=SoftFail smtp.mailfrom=attacker.com; dmarc=fail action=none header.from=trusted-brand.com; — Excerpt from an Exchange Online header showing an explicit DMARC policy bypass.
The action equals none or disposition equals none tag sitting next to a dmarc equals fail result is the smoking gun. It means the evaluating gateway correctly calculated the DMARC failure based on the lack of SPF and DKIM alignment, successfully parsed the p=reject policy from external DNS, and then intentionally decided not to enforce it. The gateway actively chose to let the message through.
This disposition tells you exactly what to look for next. The failure was not a DNS lookup timeout or a temporary parsing error. A business logic rule residing on that specific mail server intentionally overrode the standard authentication pipeline.
The Secure Email Gateway Trap
Most organizations stack a Secure Email Gateway in front of their primary cloud mailboxes. The gateway handles the heavy lifting for inbound DMARC evaluation. If your edge appliance stamped the action equals none disposition, you need to audit its policy layers. These appliances operate on a rigid hierarchy of rules where administrator-defined settings almost always take precedence over protocol-level authentication checks.
The Sins of Legacy Allow-Lists
If an IT administrator added a vendor IP address to a global allow-list three years ago to bypass a spam filter, that rule often grants a blanket pardon to any message originating from that address space. IP-based allow-lists are incredibly dangerous because cloud hosting providers constantly recycle addresses and attackers compromise legitimate marketing infrastructure. When a threat actor routes a spoofed phishing campaign through that trusted IP, the gateway immediately halts its authentication checks and delivers the payload.
The same vulnerability applies to sender domain allow-lists. A rule stating that anything from a trusted partner domain should bypass filtering is a catastrophic configuration flaw. When attackers spoof that exact partner domain, the gateway sees the allowed domain in the Header From address, triggers the exception, and completely ignores the ensuing DMARC failure. The gateway assumes the administrator knows better than the protocol.
Exchange Online and Google Workspace Overrides
If the upstream gateway correctly stamped a rejection header but the message still reached the inbox, the failure happened at the next routing hop. Microsoft 365 and Google Workspace possess complex internal routing architectures that can easily override upstream security verdicts. Administrators frequently build custom rules to ensure specific executive communications or automated alerts are never delayed.
Auditing Exchange Transport Rules
In Exchange Online, Mail Flow Rules are notorious for creating unintended security blind spots. A frequent anti-pattern involves a rule designed to prevent a specific third-party application from landing in the Junk folder. Administrators often configure these rules to set the Spam Confidence Level to negative one for those specific senders.
Setting a Spam Confidence Level to negative one does not just bypass spam filtering. It instructs Exchange to bypass almost all tenant-level security evaluations for that message. If an attacker spoofs the criteria matching that transport rule, Microsoft will deliver the message straight to the inbox regardless of the upstream DMARC failure or any malicious attachments. You must interrogate the X-MS-Exchange-Organization-AuthAs header to determine if an internal rule granted this elevated trust.
Advanced Delivery and Routing Anomalies
Both major cloud providers offer dedicated configurations designed explicitly for third-party phishing simulation tools. These advanced delivery policies intentionally blind the security stack to allow simulated attacks to reach users. If an attacker manages to hijack an IP space belonging to a sanctioned phishing simulation vendor or discovers the specific header keys used by those tools, they inherit that complete security bypass and sail past your enforcement policies.
The Remediation Playbook
Closing these gaps requires replacing broad implicit trust with granular explicit trust. You cannot simply delete every allow-list in the organization without breaking legitimate mail flow, but you can systematically narrow the blast radius. Begin by exporting your entire Mail Flow Rule configuration and identifying any rule that bypasses filtering based solely on a sender domain or an unverified sender address.
IP-based allow-lists must be deprecated entirely. The era of static IP trust is over. If a sender is struggling with deliverability, they need to fix their SPF records or correct their DKIM signing practices rather than relying on your organization to punch a hole in your defensive perimeter. You are doing them no favors by masking their poor infrastructure management.
If you absolutely must maintain a domain-based exception for a broken third-party sender, scope the rule tightly. Bind the exception to a specific sender IP, require TLS enforcement via MTA-STS, and set a hard expiration date on the rule so it does not transform into permanent technical debt. Security exceptions should be treated like temporary firewall rules, requiring constant justification and eventual removal.
Handling ARC Chain Complications
Sometimes the override is not local at all, but rather inherited from a trusted intermediary. The Authenticated Received Chain defined in RFC 8617 allows forwarders and mailing lists to preserve the original authentication results before they modify the message headers or body. If your environment is configured to trust a specific forwarder, your mail server might accept their ARC-Seal as valid.
If an attacker compromises an account at that trusted forwarder and uses it to blast out spoofed messages, your infrastructure might honor the ARC chain over the direct DMARC failure. Investigating this requires extracting the arc-pass verdicts from the headers and verifying whether your tenant settings explicitly list that forwarder as a trusted ARC sealer. If they do, the forwarder effectively laundered the malicious email on behalf of the attacker.
The takeaway
Email authentication is a continuous chain of custody. A reject directive from a sender is a powerful tool, but it functions entirely on the honor system. If your local routing rules prioritize administrative convenience over cryptographic authentication, attackers will map out those internal exceptions and abuse them to deliver payloads directly to your users.
Finding these misconfigurations before an incident occurs requires continuous visibility into your routing architecture. Analyzing raw headers manually works for isolated triage, but utilizing a platform like MailSleuth.AI helps aggregate those authentication dispositions across your entire environment. This visibility allows SOC analysts to pinpoint exactly which gateways, routing rules, or legacy transport policies are silently overriding your security posture at scale.
We dissect phishing campaigns and email infrastructure so you don't have to.


