Introduction
A critical flaw in the FortiOS OSPF daemon may let an unauthenticated attacker on an adjacent network cause a FortiGate to broadcast Link State Advertisements, the messages OSPF routers use to build their shared picture of the network. For organizations running OSPF on Fortinet firewalls, that means a host on a reachable segment could get routing information accepted by a routing domain that is supposed to trust only authenticated neighbors.
Fortinet tracks the issue as CVE-2026-126253 in PSIRT advisory FG-IR-26-267, published November 10, 2026, with a CNA score of 9.3 (Critical). NVD lists publication on November 11, 2026. The supplied CVE record lists the vendor as Unknown, but the affected vendor is Fortinet. The flaw affects FortiOS 7.2 through 8.0 and several FortiSASE releases, and Fortinet reports no known exploitation.
Technical information
Root cause: CWE 305 in the OSPF daemon
Fortinet classifies the weakness as CWE 305, Authentication Bypass by Primary Weakness, located in the OSPF daemon of FortiOS and FortiSASE. MITRE defines CWE 305 as a case where the authentication algorithm is sound, but the implemented mechanism can be bypassed because of a separate primary weakness. The classification describes the shape of the bug. It does not identify the specific programming error, and Fortinet has not disclosed that error.
The authentication being bypassed is OSPF neighbor authentication: plain text or MD5 keyed digests on OSPF packets, which exist to stop rogue routers from injecting routes. According to the advisory, a specially crafted request lets an unauthenticated attacker skip that check and still cause LSAs to be flooded to the routing domain. Mallory's analysis describes the result as a bypass of OSPF authentication that allows forged routing advertisements from an adjacent network.
The advisory metadata labels the component as "API", but the vulnerability description specifically identifies the OSPF daemon and does not describe an HTTP endpoint.
Why unauthorized LSAs matter
OSPF (Open Shortest Path First) routers exchange LSAs (Link State Advertisements). RFC 2328 describes LSAs as carrying routing information, flooded within their applicable area or autonomous system scope, and used by OSPF to calculate routes. An LSA that the routing domain accepts therefore feeds directly into route selection. Red Tiger Security interprets the potential outcome as false routing information that could redirect or interrupt traffic, and notes that the issue matters where OSPF is enabled. Fortinet's own description is narrower: unauthorized LSA broadcast.
Severity and documented attack conditions
Fortinet's CNA vector is AV:A/AC:L/PR:N/UI:N/S:C/C:N/I:H/A:H:
- Attack vector: Adjacent (AV:A)
- Attack complexity: Low (AC:L)
- Privileges required: None (PR:N)
- User interaction: None (UI:N)
- Scope: Changed (S:C)
- Confidentiality: None (C:N)
- Integrity: High (I:H)
- Availability: High (A:H)
NVD had not provided a separate assessment in the retrieved record. The critical score does not mean unrestricted internet reachability. The attacker needs adjacent network access to a segment where the OSPF daemon is reachable. There is also no confidentiality impact, so the risk is about routing integrity and service continuity, not confirmed data theft or code execution.
Attack flow
Based on the advisory, the CVSS vector and the CWE class, exploitation would look like this at a high level:
- The attacker gains a position on a network segment adjacent to a vulnerable FortiGate where the OSPF daemon is listening. Detection guidance from InfoSec Monkey identifies OSPF enabled on guest, DMZ or lab VLANs, or over IPsec to partner sites, as the highest-risk cases.
- The attacker sends a specially crafted request to the OSPF daemon. No credentials and no user interaction are required.
- Because of the primary weakness, the OSPF authentication decision is bypassed rather than failed.
- The daemon accepts the input and LSAs are broadcast into the routing domain.
- Other routers incorporate the advertised information into route calculation, which, per Red Tiger Security, could redirect or interrupt traffic.
Fortinet has not published a packet-level reproduction, and no verified public proof of concept exists. FortiOS is closed source, so there is no public commit or diff. Treat anything claiming to show the "fix diff" with suspicion.
What the fix has to do
Fortinet has not published internals, but the bug class implies what a fixed ospfd must do: make the authentication decision before any input can lead to LSA installation or flooding, with no code path that reaches flooding without that check. This is an inference from the CWE class and the advisory wording, not a confirmed implementation detail.
Patch information
Fortinet shipped firmware updates under advisory FG-IR-26-267. Fixed releases by branch:
- FortiOS 8.0: 8.0.0 through 8.0.1 vulnerable, fixed in 8.0.2 or above
- FortiOS 7.6: 7.6.0 through 7.6.7 vulnerable, fixed in 7.6.8 or above
- FortiOS 7.4: 7.4.0 through 7.4.12 vulnerable, fixed in 7.4.13 or above
- FortiOS 7.2: 7.2.0 through 7.2.13 vulnerable, fixed in 7.2.14 or above
- FortiSASE 26.3: 26.3.1 and 26.3.1_a vulnerable, fixed by Fortinet in 26.3.2. The service is cloud managed, so no customer action is required
- FortiSASE 26.2, 26.1, 25.4, 25.3, 25.2, 25.1.a, 25.1: all versions vulnerable. Migrate to a fixed release
- FortiSASE 24.4 and earlier (back to 22.4): not affected
The fix was backported into every supported FortiOS train from 7.2 up, so you can stay on your current branch and move to its next maintenance build rather than jumping majors.
Upgrade path matters
Fortinet directs customers to plan the hop with its Upgrade Path Tool. Some FortiGate upgrades need intermediate builds, and skipping them can cause configuration migration problems. The tool takes product and model, current firmware and target firmware. Its documentation says to:
- Review model-specific notes
- Back up the configuration and license file
- Read the release notes for every hop, since prerequisites may apply
- Follow recommended hops in order and allow services to stabilize between hops
- Use the documented procedure for high-availability clusters
A security target such as 7.4.13 is the destination, not permission to skip intermediate installations.
Other patch facts
- Virtual patch: no IPS signature or virtual patch covers the flaw, so firmware is the only vendor fix.
- Discovery: Loic Pantano of Fortinet's Product Security team found the issue internally, with Fortinet crediting the AI model "Claude Mythos" as an aid. No external report or known exploitation triggered the fix.
- Verification: after upgrading, confirm the build is at or above the fixed version for your branch:
get system status
Third-party scanners such as Tenable's FortiGate plugin for FG-IR-26-267 check the same thresholds (7.2.14, 7.4.13, 7.6.8, 8.0.2).
Interim measures
Red Tiger Security recommends prioritizing firewalls running OSPF and, if patching must wait, restricting OSPF to known neighbors on trusted interfaces and not enabling OSPF on internet-facing or business-facing ports. These are secondary-source recommendations, not a Fortinet-certified workaround, and they do not make an unpatched device safe. Fortinet's advisory supplies no workaround commands.
Detection methods
No vendor signatures or community detection rules exist for this CVE yet. CVETodo's coverage tracker reported no Sigma, Suricata, YARA or Nuclei rules as of November 11, 2026. One upstream system claimed that nine KQL/Sigma rules and eleven IoCs exist, but it cited no source and nothing we fetched supports it. Treat that claim as unverified. Practical detection comes down to finding vulnerable builds and watching OSPF behavior on the FortiGate.
1. Find vulnerable builds
Tenable's Nessus Plugin ID 378603 (fortigate_FG-IR-26-267.nasl, Firewalls family) is a local, authenticated check using the KB items Host/Fortigate/model and Host/Fortigate/version. It flags 7.2.x before 7.2.14, 7.4.x before 7.4.13, 7.6.x before 7.6.8 and 8.0.x before 8.0.2. It reads the reported version only and does not test the flaw, so a hit means "vulnerable build", not "exploited".
Exposure also depends on whether OSPF is running and which segments can reach it. A vulnerable FortiGate with OSPF disabled, or limited to trusted point-to-point links, carries much less risk than one with OSPF on guest, DMZ or lab VLANs, or over IPsec to partner sites. Check those first.
CVETodo's copy of the CVE record lists FortiOS 7.0.1 through 7.0.19 as affected, while Fortinet's advisory starts at 7.2. Confirm how your scanner handles 7.0.x before treating those devices as clean.
2. Hunt for route injection
Evidence shows up in OSPF state and traffic, not in admin logins or web logs. InfoSec Monkey's countermeasure guide gives the relevant FortiOS commands.
Check the link-state database for unknown advertising routers. An injected LSA must name an advertising router. Compare the list against your known OSPF speakers and investigate any router ID you cannot map to a real device. Take a baseline now so later changes stand out.
get router info ospf database brief
Check adjacency health. Every expected neighbor should be in Full state. Unknown neighbors or flapping adjacencies deserve a closer look, since forged LSAs can cause routing instability.
get router info ospf neighbor
Capture OSPF traffic (IP protocol 89). Run this on interfaces that should never carry OSPF, such as guest, DMZ or user VLANs, where any protocol 89 packet is suspicious. On real OSPF interfaces, look for sources not on your neighbor list.
diagnose sniffer packet port3 'proto 89' 4 10
CVETodo's guidance is similar: watch for unusual broadcasts or LSA traffic as a sign of exploitation attempts.
3. Use local-in policy logs as a tripwire
If you have local-in policies that accept protocol 89 only from known neighbors and deny everything else, the hit counter on the deny rule should only climb when something other than a legitimate peer talks to the port. List active rules and their order with:
show firewall local-in-policy
Alert on hits against the OSPF deny rule. On a well-understood network, each one is a good lead.
4. Do not rely on authentication failure events
Because the flaw is CWE 305, InfoSec Monkey warns that the OSPF authentication step itself is likely what gets bypassed, although Fortinet's summary does not confirm this. Successful exploitation may produce no MD5 or authentication failure events. Base detection on link-state database contents and packet sources, not auth logs.
As of publication there are no public IoCs, packet signatures or exploit artifacts to match against.
Affected systems and versions
FortiOS (per Fortinet advisory FG-IR-26-267):
- FortiOS 8.0.0 through 8.0.1
- FortiOS 7.6.0 through 7.6.7
- FortiOS 7.4.0 through 7.4.12
- FortiOS 7.2.0 through 7.2.13
FortiSASE:
- 26.3.1 and 26.3.1_a (remediated by Fortinet in 26.3.2)
- All versions of 26.2, 26.1, 25.4, 25.3, 25.2, 25.1.a and 25.1 (migration required)
- Not affected: 24.4, 24.3, 24.1, 23.4, 23.3, 23.2, 23.1 and 22.4
Unresolved discrepancy: NVD's affected configuration additionally lists FortiOS 7.0.1 through 7.0.19, but its description and Fortinet's remediation matrix omit that branch, and no fixed 7.0 release is identified. The status of 7.0.x cannot be resolved from current sources.
Vulnerable configuration: the flaw is in the OSPF daemon and requires adjacent network access, so practical exposure centers on devices with OSPF enabled, especially on untrusted or shared segments.
Vendor security history
FortiOS has prior exploited vulnerabilities with lasting consequences. On April 10, 2025, Fortinet published an investigation into activity involving previously known flaws that the Canadian Centre for Cyber Security identifies as CVE-2022-42475, CVE-2023-27997 and CVE-2024-21762. An actor created a symbolic link between user and root filesystems in a directory used to serve SSL VPN language files. That link could survive patching of the original vulnerability and allow continued read-only access to files, potentially including configurations. Fortinet responded with customer communications, an AV/IPS signature to detect and remove the link, and software changes to remove it and stop the SSL VPN service from serving malicious symbolic links.
That case showed that closing initial access does not automatically remove persistence. It concerns different vulnerabilities and a different component, though, and it is not evidence that CVE-2026-126253 provides filesystem access or has been exploited.
References
- NVD: CVE-2026-126253
- CVE Record: CVE-2026-126253
- NVD API record for CVE-2026-126253
- Fortinet PSIRT Advisory FG-IR-26-267
- Mallory: CVE-2026-126253
- InfoSec Monkey: Fortinet PSIRT Countermeasures
- CVETodo: CVE-2026-126253
- Tenable Nessus Plugin 378603
- Fortinet Upgrade Path Tool for FortiGate / FortiOS
- CIS Advisory: Multiple Vulnerabilities in Fortinet Products
- Health ISAC Bulletin: Multiple Vulnerabilities in Fortinet Products
- HKCERT: Fortinet Products Multiple Vulnerabilities
- Red Tiger Security: Fortinet November 2026 PSIRT
- MITRE CWE 305: Authentication Bypass by Primary Weakness
- RFC 2328: OSPF Version 2
- Fortinet Blog: Analysis of Threat Actor Activity
- Canadian Centre for Cyber Security: Compromise and persistent access of Fortinet FortiOS products
- CISA: Fortinet Releases Security Advisories for FortiOS (2024)
- Fortinet: About Us
- Fortinet: 2026 Gartner Magic Quadrant for Hybrid Mesh Firewall announcement



