It's like leaving a VIP badge printer set to 'print for anyone' — except the printer is off by default and someone had to deliberately plug it in
CVE-2026-26035 is an authentication bypass in Fortinet FortiWeb (versions 7.0.0–7.0.12, 7.2.0–7.2.12, 7.4.0–7.4.11, 7.6.0–7.6.6, and 8.0.0–8.0.2) that lets an unauthenticated attacker log into the management GUI or CLI with *any* username and password. The flaw lives in the RADIUS-based admin authentication flow: when an admin account is configured as Remote Type with the wildcard option enabled, the appliance's credential-verification logic is effectively short-circuited. The attacker walks straight into a full admin session — no brute-force, no token theft, just arbitrary credentials accepted.
The NVD-style CVSS vector (AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H → 9.8) treats this as a zero-click, zero-auth, internet-facing RCE-equivalent. That overstates reality. Fortinet themselves rate it High (roughly 8.8) because the wildcard setting is off by default and requires a deliberate, non-standard admin configuration. In noisgate's assessment the vendor is closer to correct: the vulnerable population is a fraction of the installed base, internet exposure is minimal (~300 FortiWeb management interfaces on Shodan), there is no public PoC, no KEV listing, and EPSS sits at the 0.51-percentile mark. However, FortiWeb is a network-edge WAF — compromise grants traffic inspection, policy manipulation, and a pivot point behind the perimeter — so the role-multiplier keeps the floor at HIGH.
4 steps from start to impact.
Identify FortiWeb management interface
Server: FortiWeb) and the /fwb/login path. Shodan and Censys tags make this trivial. Approximately 300 non-honeypot FortiWeb management interfaces are directly internet-exposed.- Network reachability to FortiWeb management port (443/tcp or 8443/tcp)
- Best practice is to restrict management to an out-of-band network or IP allowlist — most enterprises do not expose the GUI to the internet
- Only ~300 instances internet-exposed per Shodan (after honeypot filtering)
product:FortiWeb; Censys query services.software.product=FortiWebConfirm RADIUS wildcard admin config
- Target FortiWeb must have a Remote RADIUS admin account configured
- The wildcard setting must be enabled on that account (non-default)
- Wildcard is off by default — an admin must deliberately enable it
- Most FortiWeb deployments use local accounts or LDAP, not RADIUS wildcard
- Fortinet's own documentation treats wildcard as a convenience shortcut, not a recommended pattern
Obtain full administrative access
- Successful login from step 2
- Session logging is on by default and will record the new admin session
- SIEM alerts on admin login from unexpected IPs will fire if configured
admin-login; SIEM correlation on new admin source IPPivot or exfiltrate
- Maintained admin session
- Network adjacency to backend application servers (inherent to WAF placement)
- Network segmentation between DMZ and internal tiers limits lateral movement
- Egress filtering on the WAF management VLAN may block C2
The supporting signals.
| In-the-wild exploitation | Not observed. Fortinet confirms no known active exploitation as of 2026-08-12. No GreyNoise tags or CISA alerts. |
|---|---|
| Proof-of-concept | None public. No PoC on GitHub, Packet Storm, or researcher blogs as of 2026-08-14. Exploitation is trivial (send any creds) but requires the non-default config to be in place. |
| EPSS | 0.0051 (0.51st percentile) — very low predicted exploitation probability in the next 30 days. |
| KEV status | Not listed on CISA KEV as of 2026-08-14. |
| CVSS vector | CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H → 9.8. The vector assumes zero friction, which overstates real-world conditions given the non-default config prerequisite. |
| Affected versions | FortiWeb 7.0.0–7.0.12, 7.2.0–7.2.12, 7.4.0–7.4.11, 7.6.0–7.6.6, 8.0.0–8.0.2. Spans 5 major branches. |
| Fixed versions | FortiWeb 8.0.3, 7.6.7, 7.4.12, 7.2.13. No fix announced for 7.0.x — upgrade required. |
| Exposure data | ~300 FortiWeb management interfaces internet-exposed per Shodan (honeypots filtered). CyCognito notes 18.2% of classified exposed orgs are in financial services. |
| Disclosure date | 2026-08-12 via Fortinet PSIRT advisory FG-IR-26-158. |
| Vendor advisory | Fortinet rates this High (not Critical) in their own advisory, citing the non-default wildcard prerequisite. |
noisgate verdict.
The single most decisive factor is the non-default configuration prerequisite: the RADIUS wildcard admin setting must be deliberately enabled, which narrows the exploitable population to a small fraction of FortiWeb deployments. Despite this friction, FortiWeb is a network-edge WAF appliance whose compromise yields traffic interception, TLS key exfiltration, and a DMZ pivot point — the role-multiplier floors the verdict at HIGH.
Why this verdict
- Non-default config prerequisite narrows the blast radius. The wildcard setting on RADIUS admin accounts is off by default. Administrators must deliberately enable it. This is not a universally exploitable condition — it gates the entire chain.
- Minimal internet exposure. Only ~300 FortiWeb management interfaces are internet-reachable per Shodan. Most enterprises restrict WAF management to internal or OOB networks, further shrinking the attack surface.
- Role multiplier: network-edge WAF appliance. FortiWeb sits inline or in reverse-proxy mode in front of web applications. Admin compromise grants WAF policy manipulation, TLS private key export, decrypted traffic inspection, and a pivot point from DMZ into backend tiers. This is a canonically high-value network-edge role, which floors the verdict at HIGH despite the configuration friction.
- No exploitation signals. Zero PoCs, zero KEV, EPSS at 0.51% — the threat-intelligence picture is quiet. This permits a downgrade from the raw 9.8 CVSS but does not justify going below HIGH given the role.
Why not higher?
A CRITICAL rating would require either active exploitation or a universally exploitable condition. Here, the wildcard RADIUS config is non-default and likely affects a single-digit percentage of FortiWeb deployments. The ~300 internet-exposed instances and absence of any PoC or KEV listing mean the urgency does not reach CRITICAL tier. If a public PoC emerges or KEV listing occurs, this should be immediately re-evaluated.
Why not lower?
FortiWeb is a network-edge security appliance — it is not a workstation or low-value host. Compromise of a WAF grants the attacker a privileged network position with access to decrypted traffic, TLS keys, and backend server connectivity. The authentication bypass itself is trivially exploitable (any credentials work) once the prerequisite config exists. Dropping to MEDIUM would ignore the catastrophic blast radius that a WAF takeover represents in any enterprise with sensitive web applications behind it.
What to do — in priority order.
- Disable the wildcard setting on all Remote-Type RADIUS admin accounts immediately — This is the vendor-recommended workaround and fully neutralizes the vulnerability without patching. Audit via CLI:
config system admin-user→ check forwildcard enable. Deploy within the noisgate mitigation SLA of 30 days for HIGH — but given the simplicity of the workaround, aim for same-day. - Restrict management interface access to allowlisted IPs or an out-of-band management network — Even if the wildcard config exists, limiting who can reach the login page eliminates remote exploitation. Use FortiWeb's trusted-host configuration or upstream ACLs. Deploy within 30 days per noisgate mitigation SLA.
- Enable and monitor admin login alerts in your SIEM — FortiWeb logs all admin authentication events. Create a detection rule for admin logins from unexpected source IPs or with previously unseen usernames. This provides detection-in-depth while patching is scheduled.
- Rotate TLS certificates and private keys on affected FortiWeb appliances after patching — If exploitation cannot be ruled out, assume TLS private keys stored on the WAF may have been exfiltrated. Re-key after patching to close the exposure window.
- **WAF signatures / IPS rules on a *different* device in front of FortiWeb** — the exploit is a normal HTTPS login POST with valid-looking form fields. There is no malformed payload for an upstream IPS to match on.
- MFA on backend applications — MFA on the apps behind the WAF does not protect the FortiWeb management plane itself. The bypass is on the WAF admin login, not application auth.
- Rate-limiting login attempts — the exploit succeeds on the *first* attempt with any credentials. Rate-limiting is irrelevant.
Crowdsourced verification payload.
Run this script on each FortiWeb appliance via SSH as a read-only admin (or higher). Usage: bash check_cve_2026_26035.sh. No arguments needed. The script inspects the running config for RADIUS wildcard admin accounts and checks the firmware version.
#!/bin/bash
# check_cve_2026_26035.sh
# Checks FortiWeb for CVE-2026-26035 exposure
# Run on the FortiWeb appliance via SSH as admin
# Exit codes: 0=PATCHED, 1=VULNERABLE, 2=UNKNOWN
set -euo pipefail
# Get firmware version
VERSION=$(fnsysctl cat /proc/version 2>/dev/null || echo "")
if [ -z "$VERSION" ]; then
# Try alternative method
VERSION=$(diagnose system info 2>/dev/null | grep -i 'version' | head -1 || echo "")
fi
if [ -z "$VERSION" ]; then
echo "UNKNOWN - Could not determine FortiWeb version. Run manually: 'get system status'"
exit 2
fi
echo "[*] FortiWeb version info: $VERSION"
# Check for wildcard admin accounts via config dump
# This requires CLI access; adapt if using API
WILDCARD_CHECK=$(show system admin-user 2>/dev/null | grep -A5 'set type' | grep -c 'set wildcard enable' || echo "0")
if [ "$WILDCARD_CHECK" -gt 0 ]; then
echo "VULNERABLE - Found $WILDCARD_CHECK admin account(s) with wildcard enabled."
echo "Remediation: Disable wildcard on Remote-Type RADIUS admin accounts or upgrade."
exit 1
else
echo "PATCHED - No admin accounts with wildcard setting enabled found."
echo "Note: Verify firmware is also updated to 8.0.3+, 7.6.7+, 7.4.12+, or 7.2.13+."
exit 0
fiIf you remember one thing.
set wildcard disable) that fully neutralizes the vulnerability and should be treated as your noisgate mitigation SLA action (within 30 days for HIGH, but aim for same-day given the simplicity). Second, restrict FortiWeb management interfaces to allowlisted IPs if you haven't already. Third, schedule firmware upgrades to FortiWeb 8.0.3, 7.6.7, 7.4.12, or 7.2.13 within the noisgate remediation SLA of 180 days. If you're still running 7.0.x, plan a branch upgrade — no patch is available for that train. Monitor Fortinet PSIRT and CISA KEV for any escalation; if a PoC drops or KEV listing appears, compress the remediation window to days, not months.Sources
What defenders are saying.
Crowdsourced verification outputs.
Results submitted by users who ran the verification payload against their environment.