Introduction
Fortinet's advisory FG-IR-26-214 says a single crafted request to the FortiMail Exchange Integration API may let a remote attacker execute unauthorized code or commands, with no login required. FortiMail sits directly in the email path of the organizations that deploy it. A flaw in this component that is reachable without credentials deserves prompt attention, even though Fortinet reports no known exploitation.
The CVE data lists the vendor as Unknown, but the affected vendor is Fortinet. FortiMail is Fortinet's secure email product. It covers inbound threat protection and outbound data loss or compliance risks, and it ships as physical appliances, virtual appliances, and a service-based deployment positioned alongside Microsoft Exchange and cloud email services. Fortinet reports that Gartner positioned FortiMail as a Challenger in the 2025 Magic Quadrant for Email Security (Gartner describes its research as opinion, not endorsement), which places it among the established secure email gateway options in enterprise environments.
Technical information
Root cause
Fortinet places the flaw in the FortiMail Exchange Integration feature and labels the affected component API. The assigned weakness is CWE-91, XML Injection, which MITRE defines as improper neutralization of special XML elements, allowing an attacker to modify XML syntax, content, or commands before the application processes them.
In practical terms, request data that reaches the Exchange Integration code is used to build XML structures or XPath expressions without adequate neutralization. An attacker who controls that data can inject XML or XPath syntax and change the meaning of the resulting document or query. Fortinet states that this may lead to unauthorized code or command execution. The description narrows the scope to the code that turns request data into XML or XPath for the Exchange integration, rather than the mail filtering engine as a whole.
Fortinet has not published vulnerable code, and FortiMail is closed source, so we have no code snippet to show.
XML injection versus blind XPath injection
The description uses both "XML injection" and "blind XPath injection". The two terms are related but not identical:
- OWASP describes blind XPath injection as a technique for extracting information when an application incorporates user-supplied data unsafely into XPath queries. When the XML structure is unknown or errors are suppressed, the attacker infers information through repeated true or false queries.
- MITRE lists blind XPath injection as an alternate term under CWE-91, but notes it is more closely associated with CWE-643, improper neutralization of data within XPath expressions.
The terminology alone does not tell us whether the FortiMail flaw is exploited through boolean inference, authentication bypass, XML restructuring, or another route to command execution. The general attack class is well understood; the specific chain for this CVE is not public.
CVSS breakdown
NVD publishes the following Fortinet-supplied base vector, scored 8.7 (High):
CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:C/C:H/I:H/A:N
- Attack Vector: Network. The component is reachable over the network.
- Attack Complexity: High. The score does not treat this as a trivial attack.
- Privileges Required: None. No account is needed.
- User Interaction: None. No victim action is needed.
- Scope: Changed. In the scoring model, impact extends beyond the vulnerable component.
- Confidentiality: High.
- Integrity: High.
- Availability: None. That is the assigned scoring value. It is not a guarantee about every operational consequence.
The advisory also lists temporal values E:U/RL:O/RC:R, meaning Unproven exploit code maturity, Official Fix available, and Reasonable report confidence. These are separate from the base metrics.
Attack flow
Based on the advisory and the weakness class, the attack proceeds at a high level as follows:
- The attacker identifies a FortiMail appliance running an affected 7.6 or 8.0 build with the Exchange Integration API reachable over the network.
- Without authenticating, the attacker sends a specially crafted request to that API containing XML or XPath syntax in a field the application uses to build XML data or an XPath query.
- Because the input is not properly neutralized, the injected syntax alters the structure or logic of the XML document or XPath expression. In a blind XPath scenario, this can involve repeated true or false queries to infer data.
- According to Fortinet, successful exploitation may result in execution of unauthorized code or commands.
The high attack complexity rating indicates that conditions beyond simply sending one request may be required. Fortinet has not disclosed the exact request path, parameter, HTTP method, payload format, or the privilege level under which commands would execute. We do not speculate beyond that, and readers should not assume root access or guaranteed appliance takeover from the public description.
Patch information
Fortinet fixed CVE-2026-126162 in firmware updates. FortiMail is closed source, so there is no public source code commit or diff to review. This section is based on Fortinet's PSIRT advisory FG-IR-26-214, which NVD lists as the CNA's official reference.
Fixed releases
- FortiMail 8.0: builds 8.0.0 through 8.0.1 are vulnerable. Fixed in 8.0.2 or later.
- FortiMail 7.6: builds 7.6.0 through 7.6.6 are vulnerable. Fixed in 7.6.7 or later.
- FortiMail 7.4: not affected, and no action is needed.
The 7.4 entry suggests the vulnerable Exchange Integration code first shipped in 7.6, which is why only the 7.6 and 8.0 lines needed patched builds.
FortiMail Cloud
Fortinet patched its own hosted service before announcing the bug. The advisory says every FortiMail Cloud instance was updated to v7.6.7 GA or v8.0.2 GA on October 5, 2026. Public disclosure followed about five weeks later, on November 10, 2026 (NVD records publication on November 11, 2026). Cloud customers do not need to do anything. Customers running their own appliances or VMs must upgrade themselves.
No virtual patch
The advisory lists "Virtual Patch: No". Fortinet has not released an IPS signature or FortiGuard Virtual Patching rule that blocks the crafted request in front of an unpatched device, and it publishes no workaround. The firmware upgrade is the only vendor fix.
What the fix probably changed (our inference)
Fortinet has not published code-level details, so this is educated inference. The standard fix for CWE-91 is to stop building XML or XPath strings by concatenating raw request input. Instead, the code uses parameterized XPath queries or variable binding, or escapes XML metacharacters such as ', ", <, >, & and /. Strict allow-list checks on fields that should only hold simple values, such as mailbox names, GUIDs, or hostnames, also help. MITRE similarly recommends an "accept known good" validation strategy, and CAPEC recommends validating XPath input, escaping, and parameterization. We would expect 7.6.7 and 8.0.2 to handle Exchange Integration input this way, but Fortinet has not confirmed it.
Discovery
The advisory credits Gwendal Guegniaud of Fortinet's Product Security team, working with the AI model Claude Mythos. Because the bug was found internally, Fortinet could ship fixed GA builds and patch its cloud fleet before going public, so fixes were available on disclosure day.
Upgrade precautions from the release notes
The FortiMail 8.0.2 and 7.6.7 release notes list these operational steps:
- Before upgrading, back up the configuration and any other stored data.
- After upgrading, clear the browser cache before logging into the web interface, and verify that the version and build number match the loaded image.
- Update antivirus signatures immediately after the image upgrade, since bundled signatures may be older than those on Fortinet's distribution network. This step is general upgrade hygiene and is not part of the XML injection fix.
To confirm you are patched, check that each appliance reports 8.0.2 or later on the 8.0 branch, or 7.6.7 or later on the 7.6 branch. Firmware is available from the Fortinet Support portal.
Affected systems and versions
- FortiMail 8.0.0 through 8.0.1 (customer-managed appliances and VMs): affected.
- FortiMail 7.6.0 through 7.6.6 (customer-managed appliances and VMs): affected.
- FortiMail 7.4: not affected.
- FortiMail Cloud: remediated by Fortinet on October 5, 2026. No customer action is required.
The vulnerable functionality is the Exchange Integration API. The advisory does not describe additional configuration preconditions, so treat any affected build as in scope for the upgrade.
Vendor security history
FortiMail has a documented exploitation history, but for a different flaw. On October 1, 2026, CISA added CVE-2026-104286, a FortiMail path traversal vulnerability, to its Known Exploited Vulnerabilities Catalog based on evidence of active exploitation. That is a separate vulnerability with a separate mechanism, and it does not establish that CVE-2026-126162 is being exploited.
Across the broader Fortinet portfolio, Mandiant reported exploitation of FortiOS SSL VPN vulnerability CVE-2022-42475 as early as October 2022, including BOLDMOVE malware with a Linux variant deployed on Fortinet devices. Targets included a European government entity and an African managed service provider, and Mandiant assessed a China nexus with low confidence. None of this provides attribution or indicators for the FortiMail flaw discussed here.
For CVE-2026-126162, Fortinet reports "Known exploited: No". No named actor, public proof of concept, indicators of compromise, or detection signature has been published, and KEV inclusion is not confirmed in the material we reviewed. Fortinet's PSIRT policy describes a dedicated team for receiving and investigating vulnerabilities, coordinated remediation, monthly advisories, and accelerated publication when active exploitation raises customer risk. The internal discovery and pre-disclosure cloud remediation of this CVE fit that process.
References
- Fortinet PSIRT Advisory FG-IR-26-214
- NVD entry for CVE-2026-126162
- CVE Record for CVE-2026-126162
- MITRE CWE-91: XML Injection
- MITRE CAPEC-83: XPath Injection
- OWASP: Blind XPath Injection
- FIRST CVSS v3.1 Specification
- FortiMail 8.0.2 Release Notes
- FortiMail 7.6.7 Release Notes
- CISA: CISA Adds One Known Exploited Vulnerability to Catalog (October 1, 2026)
- CISA Known Exploited Vulnerabilities Catalog
- Mandiant: Suspected Chinese Threat Actors Exploiting FortiOS Vulnerability (CVE-2022-42475)
- Fortinet Security Vulnerability Policy
- Fortinet: Proactive, Responsible Disclosure
- Fortinet: FortiMail Workspace Security
- AWS Marketplace: FortiMail Secure Email Gateway
- HKCERT: Fortinet Products Multiple Vulnerabilities
- Rescana: Fortinet November 2026 PSIRT, FortiMail and FortiManager Unauthenticated RCE



