Introduction
FortiManager and FortiAnalyzer sit at the center of many Fortinet deployments, holding centralized policy, configuration and log data for entire fleets of managed devices. CVE-2026-126177 lets an unauthenticated remote attacker use crafted SAML requests to inject arbitrary content into a signed SAML response, which weakens the core promise a SAML signature makes to whoever relies on it.
Fortinet rates the issue High, with a CNA-supplied CVSS v3.1 base score of 8.1, and has shipped fixed firmware across the 8.0, 7.6, 7.4 and 7.2 branches of FortiAnalyzer, FortiAnalyzer Cloud, FortiManager and FortiManager Cloud. The advisory, FG-IR-26-222, was first published on November 10, 2026. NVD published the CVE on November 11, 2026 at 22:17:16 UTC. Fortinet marks it Known Exploited: No, and no public exploit or proof of concept has been reported. This post is a brief summary of what the public sources say, and where they stop.
Technical information
Root cause: CWE-91 in the SAML code path
Fortinet classifies the bug as CWE-91, XML Injection (aka Blind XPath Injection), and lists the affected component as API. MITRE describes CWE-91 as a product failing to neutralize special elements used in XML, which lets an attacker modify XML syntax, content or commands before an end system processes it. Reserved characters in attacker input can introduce new data or change the structure of the surrounding document.
In this case the injection point is SAML handling. A hardening guide covering this advisory batch places FG-IR-26-222 on the SAML login for the admin GUI. Fortinet's description is specific about the outcome. Crafted SAML requests may allow attacker-supplied content to end up inside a signed SAML response.
That matters because a SAML signature tells the relying party that the signer vouches for every element inside the signed XML. If attacker-controlled data reaches the XML before it is signed, the signature then covers the attacker's content as well. The relying party has no cryptographic way to distinguish injected elements from legitimate ones.
Attack flow, as far as it is documented
The public record supports only a high-level sequence:
- The attacker reaches the SAML login exposed by the FortiManager or FortiAnalyzer HTTPS admin GUI. No credentials are required (PR:N) and no user interaction is needed (UI:N).
- The attacker sends a crafted SAML request containing XML special elements in fields the product later reuses.
- Because those elements are not neutralized, they are incorporated into the XML of the SAML response the product builds.
- The response is signed, so the injected content is carried inside a validly signed SAML message.
Several things are not public: the exact endpoint, the vulnerable request parameter, the payload, the injected XML structure, whether the appliance must play a particular SAML role, whether a specific authentication configuration is required, and whether the feature is enabled by default. Fortinet is closed source and has published no diff or write-up. The advisory also lists CVE-2026-126176 alongside this CVE. NVD confirms the SAML response injection behavior for CVE-2026-126177 specifically.
Reading the CVSS vector
Published base vector:
CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:C/C:L/I:H/A:L
- Attack vector: Network
- Attack complexity: High
- Privileges required: None
- User interaction: None
- Scope: Changed
- Confidentiality / Integrity / Availability: Low / High / Low
AC:H means exploitation is not assessed as trivial, even though it is unauthenticated. S:C with I:H reflects that the impact lands on whatever consumes the signed response, not just on the vulnerable component.
Fortinet also publishes a 7.7 temporal score with E:P/RL:O/RC:C appended. E:P is a scoring value and not a published payload. Fortinet's short impact label reads "Denial of service", which is narrower than both its narrative (signed content injection) and the High integrity metric. NVD had not supplied its own assessment at publication.
In our reading, none of the sources demonstrate authentication bypass, account takeover, private key theft, signature algorithm compromise or remote code execution for this CVE. Those may be intuitive consequences of signed content injection in general, but they are not established for this implementation.
Patch information
Fortinet has shipped an official fix as new firmware for all four affected product families, documented in FG-IR-26-222. The CNA record marks it RL:O (Remediation Level: Official Fix), and the CVE record's solutions field lists an upgrade path for each product.
Each branch gets its own maintenance release. You need the listed build in your branch, or anything later:
- 8.0 branch: vulnerable 8.0.0, first fixed build 8.0.1
- 7.6 branch: vulnerable 7.6.0 through 7.6.7, first fixed build 7.6.8
- 7.4 branch: vulnerable 7.4.0 through 7.4.11 (Cloud: 7.4.1 through 7.4.11), first fixed build 7.4.12
- 7.2 branch: vulnerable 7.2.0 through 7.2.12 (Cloud: 7.2.1 through 7.2.12), first fixed build 7.2.13
The same numbers apply to all four products. A FortiManager on 7.4.9 and a FortiAnalyzer Cloud tenant on 7.4.9 are both fixed by 7.4.12.
What the fix likely involves. Fortinet has not published the code change, so it cannot be verified. What follows is general guidance for this vulnerability class, not a confirmed description of Fortinet's changes. A fix would normally include some or all of the following:
- Treating values from the incoming request as text, never as markup, by escaping or encoding them or by building the response with a DOM or XML API instead of string concatenation.
- Validating request fields against a strict schema before they are used in the response.
- Ensuring only server-generated, trusted data is placed inside the signed assertion.
No virtual patch, no workaround. The advisory states Virtual Patch: No. There is no FortiGuard IPS signature, no CLI workaround and no SAML configuration toggle published for this issue. Installing fixed firmware is the vendor remediation. Workarounds from other advisories are not a supported substitute.
Discovery. Fortinet's Product Security team found the bug internally, credited to Loic Pantano working with the frontier AI model Claude Mythos. There is no outside researcher report and no public third-party analysis of the fix.
Verifying the fix. Because the fix is firmware, verification is a version check. An appliance is fixed when it runs 8.0.1, 7.6.8, 7.4.12 or 7.2.13 or later in its branch. For Cloud editions, confirm your tenant shows a fixed build. The CIS alert for the wider Fortinet release recommends applying vendor updates immediately after appropriate testing.
Detection methods
Today you can find vulnerable installations, but you cannot reliably detect exploitation attempts. Every public detection artifact at the time of writing is a version check, not a payload inspection.
Authenticated scanning with Nessus
Tenable has two plugins tied to FG-IR-26-222, in the Firewalls family, published 11/11/2026:
- Plugin 378654, Fortinet FortiAnalyzer XML injection (FG-IR-26-222), file
fortianalyzer_FG-IR-26-222.nasl - Plugin 378655, Fortinet FortiManager XML injection (FG-IR-26-222), file
fortimanager_FG-IR-26-222.nasl
Caveats worth knowing:
- Version only. Tenable notes that Nessus has not tested for these issues and relies on the self-reported version. A finding means the build is in range, not that it was exploited. Unusual builds or misreported versions can produce incorrect results.
- Credentialed scans required. Both plugins are
Localand depend on the KB itemsHost/Fortigate/modelandHost/Fortigate/version. An unauthenticated network scan will not produce a finding. - One finding, two CVEs. Both plugins cover CVE-2026-126176 and CVE-2026-126177, so expect one result per host.
- Feed freshness. Make sure your plugin feed has updated past 11/11/2026 before trusting a clean result.
Manual version checks
Without Nessus, pull the firmware version from each appliance and compare it against the advisory ranges. The hardening guide covering this batch uses this command on FortiManager and FortiAnalyzer:
get system status
For each appliance, answer two questions: is the build affected, and can untrusted networks reach the HTTPS admin GUI with its SAML login? Hosts where both answers are yes are the highest priority.
Network and log signatures: none published
- No IPS virtual patch. Fortinet lists Virtual Patch: No, unlike some other advisories in the same November 2026 release.
- No community rules. As of 11 November 2026, CVETodo listed no Sigma, Suricata, YARA or Nuclei detections, and no public exploit or Nuclei template.
- Generic guidance only. Watch traffic to SAML endpoints for suspicious or malformed SAML requests, and use a WAF capable of inspecting XML or SAML payloads. No source provides a concrete pattern, so treat this as hunting advice.
Hunting without a signature
- Traffic to the SAML login: unauthenticated SAML requests to the FortiManager or FortiAnalyzer admin GUI from addresses outside your identity provider and admin networks, and SAML messages that are unusually large or heavy with XML markup, if your reverse proxy or WAF logs request bodies.
- Appliance login records: SAML admin logins that do not match a real IdP-initiated session, such as unfamiliar accounts, unusual source IPs or odd times. Since the bug targets the integrity of signed SAML responses, a login that looks valid but has no matching IdP event is the anomaly to chase.
There is no published indicator of compromise. Expect version checks to remain the primary control until payload-aware signatures appear.
Affected systems and versions
All ranges are inclusive.
FortiAnalyzer
- 8.0.0 (fixed in 8.0.1 or above)
- 7.6.0 through 7.6.7 (fixed in 7.6.8 or above)
- 7.4.0 through 7.4.11 (fixed in 7.4.12 or above)
- 7.2.0 through 7.2.12 (fixed in 7.2.13 or above)
FortiAnalyzer Cloud
- 8.0.0 (fixed in 8.0.1 or above)
- 7.6.0 through 7.6.7 (fixed in 7.6.8 or above)
- 7.4.1 through 7.4.11 (fixed in 7.4.12 or above)
- 7.2.1 through 7.2.12 (fixed in 7.2.13 or above)
FortiManager
- 8.0.0 (fixed in 8.0.1 or above)
- 7.6.0 through 7.6.7 (fixed in 7.6.8 or above)
- 7.4.0 through 7.4.11 (fixed in 7.4.12 or above)
- 7.2.0 through 7.2.12 (fixed in 7.2.13 or above)
FortiManager Cloud
- 8.0.0 (fixed in 8.0.1 or above)
- 7.6.0 through 7.6.7 (fixed in 7.6.8 or above)
- 7.4.1 through 7.4.11 (fixed in 7.4.12 or above)
- 7.2.1 through 7.2.12 (fixed in 7.2.13 or above)
The one difference between deployment types is in the 7.4 and 7.2 branches, where Cloud ranges start at 7.4.1 and 7.2.1 and the appliance ranges start at 7.4.0 and 7.2.0. The relevant exposure is the SAML login on the admin GUI. The sources do not state whether a particular SAML configuration is required or whether it is enabled by default.
Vendor security history
FortiManager has a documented history of being targeted. CVE-2024-47575, a different FortiManager flaw, was mass exploited by an actor Mandiant tracks as UNC5820, with activity observed as early as June 27, 2024. That bug allowed an unauthorized, attacker-controlled FortiManager device to execute arbitrary code or commands against vulnerable FortiManager instances. Mandiant observed staging and exfiltration of managed FortiGate configuration data, including users and hashed passwords, though it found no evidence in the activity it analyzed that the data was used for lateral movement, and it could not assess the actor's motivation or location. CISA highlighted Fortinet's updated workarounds and indicators and added the CVE to its Known Exploited Vulnerabilities Catalog.
That history explains why management-plane products deserve attention, but the UNC5820 attribution belongs to CVE-2024-47575 alone. For CVE-2026-126177, the record shows an internally discovered issue, published fixes across all affected branches, and no vendor-reported exploitation. Fortinet's Gartner recognition as a Hybrid Mesh Firewall Leader is a market position, not a security assessment of FortiManager or FortiAnalyzer.
References
- Fortinet PSIRT advisory FG-IR-26-222
- NVD entry for CVE-2026-126177
- NVD API record for CVE-2026-126177
- CIRCL Vulnerability Lookup: CVE-2026-126177
- VulDB: CVE-2026-126177 in FortiAnalyzer
- CVETodo: CVE-2026-126177
- Tenable Nessus Plugin 378654 (FortiAnalyzer)
- Tenable Nessus Plugin 378655 (FortiManager)
- InfosecMonkey: Fortinet PSIRT countermeasures
- MITRE CWE-91: XML Injection (aka Blind XPath Injection)
- FIRST CVSS v3.1 Specification
- CIS: Multiple Vulnerabilities in Fortinet Products
- NHS England Digital cyber alert CC 4888
- HKCERT: Fortinet Products Multiple Vulnerabilities
- CERT FR: Multiples vulnérabilités dans les produits Fortinet
- SecurityWeek: Fortinet Patches Nearly 100 Vulnerabilities
- FortiManager Data Sheet
- FortiAnalyzer product page
- Fortinet press release: 2026 Gartner Magic Quadrant for Hybrid Mesh Firewall
- Google Cloud / Mandiant: Investigating FortiManager Zero Day Exploitation (CVE-2024-47575)
- CISA: Fortinet Updates Guidance and Indicators of Compromise following FortiManager Vulnerability Exploitation



