Two separate denial-of-service vulnerabilities in Cisco Secure Firewall remote access VPN are being exploited in the wild. Cisco published the first in August and the second in October, both carry a CVSS base score of 8.6, and both are in CISA's Known Exploited Vulnerabilities catalog. Neither has a workaround.

The detail that matters, and the reason an estate patched in September is not covered: the two advisories have different fixed releases in the same software trains. On ASA 9.16 the first is fixed in 9.16.4.103 and the second in 9.16.4.104. One digit apart, three months apart, and a maintenance window that addressed the first does nothing for the second.

What is actually being exploited

Both affect Cisco Secure Firewall Adaptive Security Appliance Software and Cisco Secure Firewall Threat Defense Software, and both target remote access VPN rather than the inspection path. The impact is availability: a successful attack crashes the device. For an organization whose remote workers arrive through that firewall, a crash is an outage.

 CVE-2026-20349CVE-2026-76418
Cisco advisory published11 August 202628 October 2026
CVSS base score8.68.6
ClassHeap inspectionUnchecked return value leading to a NULL pointer dereference
Cisco on exploitationPSIRT aware of active exploitation in August 2026PSIRT aware of active exploitation in October 2026
Added to CISA KEV11 August 202629 October 2026
Federal remediation due14 August 20261 November 2026
WorkaroundNoneNone; a mitigation exists and carries a cost

Read the two KEV rows together. Each carries a three-day remediation window for federal civilian agencies, and both dates have passed. If you are subject to that directive, you are already outside it. If you are not, the KEV listing still tells you something a CVSS score does not: this is being used, not theorized.

The fixed releases, and why one patch is not both

This is the section to take to whoever runs the change window. The trains are the same, the target releases are not.

ASA trainFixes CVE-2026-20349Fixes CVE-2026-76418
9.16 and earlier9.16.4.1039.16.4.104
9.189.18.4.949.18.4.96
9.209.20.4.499.20.4.50
9.229.22.3.269.22.3.200
9.239.23.1.479.23.1.48
9.249.24.1.269.24.1.27 or 9.24.10.1
FTD trainFixes CVE-2026-20349Fixes CVE-2026-76418
7.0 and earlier7.0.107.0.11
7.27.2.127.2.13
7.47.4.87.4.9
7.67.6.67.6.7
7.77.7.137.7.14
10.010.0.210.0.3
10.110.1.010.1.1

The later release in each row addresses both. Go to it directly rather than stepping through the earlier one, and record the version you land on against both CVE numbers, because an audit six months from now will ask for exactly that.

The mitigation, and what it costs

Cisco publishes no workaround for either vulnerability. For CVE-2026-76418 it publishes a mitigation: disable secondary authentication, either on the user-configured tunnel group list or on DefaultWEBVPNGroup, with no tunnel-group-list enable or no secondary-authentication-server-group depending on the configuration.

Cisco states plainly that both reduce the security posture of remote access VPN. Secondary authentication is a control somebody put there deliberately, and turning it off to survive a denial-of-service trades one exposure for another. Treat it as a bridge measured in days while the upgrade is scheduled, and write down the date it gets reversed. A temporary mitigation with no end date is a permanent weakening that nobody remembers agreeing to.

Which configurations are exposed

For CVE-2026-20349, Cisco lists the device as affected when it runs a vulnerable release and carries one of these configurations:

  • IKEv2 remote access VPN with client services enabled
  • SSL VPN, meaning webvpn enabled on an interface
  • Zero Trust Network Access, on Threat Defense only

Cisco confirms that Secure Firewall Management Center Software is not affected. That is worth knowing before anyone schedules an outage for a device that does not need one.

Check what is actually enabled rather than what the design document says. Remote access VPN configurations accumulate: a tunnel group stood up for a project in 2022 and never removed still counts.

If you are on 9.16 or 7.0, this is a different conversation

Both tables start with a row reading "and earlier". An ASA on 9.16 or a Threat Defense on 7.0 is at the back of its support window, and the fix for that row is the last one it will receive in that train. Patching buys the quarter. It does not change the fact that the platform is near the end of the line, and the next advisory may not have a row for it at all.

If that describes your estate, the patch is this week's work and the replacement is this cycle's budget. Those are two decisions and they want separate owners, but they want making at the same time, because a firewall that reaches end of support while carrying every remote worker in the business is not a security problem that waits politely for the next planning round.

What to do this week

  • Inventory the running version on every ASA and Threat Defense device, not the version in the asset register. Those disagree more often than anyone expects.
  • Compare against the right column. The second column covers one CVE. The third covers both.
  • Check whether remote access VPN is actually enabled on each device, using the three configurations above.
  • Schedule to the later release in the row and record the landing version against both CVE numbers.
  • If a mitigation is applied first, put a reversal date on it in the same change record.
  • Flag anything on 9.16 or 7.0 for the lifecycle conversation rather than only the patch conversation.

Common questions

Does a crash mean a breach? These two are availability vulnerabilities. Cisco rates both for impact to availability rather than to confidentiality or integrity. That does not make them minor, because a firewall crash on a remote access path is an outage for everyone arriving through it, but it is a different incident from data exfiltration and should be described accurately to leadership.

We patched in September. Are we covered? For CVE-2026-20349, if you landed on the release in the second column, yes. For CVE-2026-76418, no. That advisory did not exist in September.

Is Secure Firewall Management Center affected? Cisco confirms it is not affected by CVE-2026-20349.

We are not a federal agency. Does the KEV due date mean anything to us? It is not binding on you. It is a useful signal: CISA sets those dates by assessed risk, and a three-day window on both is the shortest category they use.

Sources

Find out which release every firewall in your estate is actually on

Trybus Solutions inventories the running version on each device, checks which remote access VPN configurations are enabled, and gives you the upgrade sequence in writing against both CVE numbers.

Or call 423-633-1817 · info@trybussolutions.com