← cd ../blog

CVE-2026-0322: PAN-OS management interface code injection to root, exploited in the wild

An authenticated code-injection flaw in the PAN-OS management interface (CVSS 4.0 8.6, on CISA KEV) gives root on firewalls and Panorama. Affected builds, exposure checks, IOCs and the fix.

CVE-2026-0322 is a code-injection bug in the PAN-OS management web interface that turns a valid admin login into root on the firewall. Palo Alto Networks found it internally, disclosed it on 14 October 2026 with “very limited” exploitation already observed, and CISA added it to KEV the same day. It needs credentials, so it isn’t a spray-the-internet bug, but a root shell on your perimeter firewall is about as bad as access gets. Related reading: the PHP2Shell post also touches PAN-OS management interfaces, and the 2026 RCE field guide has the wider picture.

Scope and ethics. This post covers exposure checks, detection and remediation for firewalls and Panorama you own or are authorized to assess. It contains no exploit details. Palo Alto hasn’t published the vulnerable code path, and neither will I.

At a glance

   
CVE CVE-2026-0322, “PAN-OS: Authenticated Remote Code Execution (RCE) in the Management Interface”
Severity CVSS 4.0 8.6 High, AV:N/AC:L/AT:N/PR:H/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N/E:A (Palo Alto PSIRT). 8.5 if management access is restricted to a jump box. NVD’s CVSS 3.1 score is 7.2. Vendor urgency: Highest
CWE CWE-94 Code Injection (CAPEC-242)
Affected PAN-OS on PA-Series and VM-Series firewalls and Panorama (virtual and M-Series): 10.2, 11.1, 11.2, 12.1, 12.2 branches. Cloud NGFW and Prisma Access aren’t affected
Fixed in 12.2.4, 12.1.10-h2, 11.2.16, 11.1.16-h3, 10.2.18-h12 (plus hotfixes on older maintenance lines, see below)
Exploited? Yes. Vendor: “very limited exploitation” (exploit maturity ATTACKED). CISA KEV added 14 Oct 2026 (federal due date 17 Oct), forensic triage required
Auth / interaction High-privilege management credentials (PR:H) plus network access to the management interface / none

Description: what we know about the bug

Palo Alto’s advisory says a code-injection flaw in the management interface lets “an authenticated attacker with network access” execute arbitrary code with root privileges. The web management plane runs with far more privilege than the admin role that’s logged into it. A code-injection bug there means admin-supplied input ends up evaluated as code by a component running as root, so the PAN-OS role model (superuser, device admin, read-only and so on) stops being the boundary.

That’s as far as public information goes. The bug was found internally (credited to Suraj Siddharudh of Palo Alto Networks), and as of this writing neither the vendor nor any research write-up I could find describes the vulnerable endpoint or parameter. Treat any third-party “root cause” claims with suspicion until the vendor confirms them. One secondary source claims a low-privilege account is enough, but the vendor’s vector says PR:H, and that’s what I’d plan around.

The “Required configuration” section of the advisory says no special configuration is needed. If the management interface (or a data-plane interface with a management profile) is reachable and the build is unpatched, the device is affected.

Impact

  • Root on the firewall or Panorama. An attacker could change policy, decrypt keys, plant persistence below the admin UI, capture traffic, and pivot into every zone the firewall touches.
  • Panorama multiplies it. Root on Panorama means a path to pushing config to every managed firewall.
  • Credential angle. PR:H means the attacker already has (or has stolen) admin credentials. A compromise suggests an earlier credential exposure you also need to investigate.

Which deployments are exposed

The full fixed-version matrix from the advisory (pick the hotfix on your maintenance line, or jump to the latest fixed release):

Branch Affected Fixed in
12.2 12.2.0 – 12.2.3 12.2.4+
12.1 12.1.8 – 12.1.10-h* 12.1.10-h2+
12.1 12.1.5 – 12.1.7-h* 12.1.7-h8, or 12.1.10-h2+
12.1 12.1.2 – 12.1.4-h* 12.1.4-h12, or 12.1.10-h2+
11.2 11.2.14 – 11.2.15 11.2.16+
11.2 11.2.11 – 11.2.13-h* 11.2.13-h3, or 11.2.16+
11.2 11.2.8 – 11.2.10-h* 11.2.10-h15, or 11.2.16+
11.2 11.2.5 – 11.2.7-h* 11.2.7-h21, or 11.2.16+
11.2 11.2.0 – 11.2.4-h* 11.2.4-h22, or 11.2.16+
11.1 11.1.14 – 11.1.16-h* 11.1.16-h3+
11.1 11.1.11 – 11.1.13-h* 11.1.13-h15, or 11.1.16-h3+
11.1 11.1.8 – 11.1.10-h* 11.1.10-h35, or 11.1.16-h3+
11.1 11.1.7 11.1.7-h11, or 11.1.16-h3+
11.1 11.1.5 – 11.1.6-h* 11.1.6-h39, or 11.1.16-h3+
11.1 11.1.0 – 11.1.4-h* 11.1.4-h37, or 11.1.16-h3+
10.2 10.2.17 – 10.2.18-h* 10.2.18-h12+
10.2 10.2.14 – 10.2.16-h* 10.2.16-h11, or 10.2.18-h12+
10.2 10.2.11 – 10.2.13-h* 10.2.13-h26, or 10.2.18-h12+
10.2 10.2.8 – 10.2.10-h* 10.2.10-h41, or 10.2.18-h12+
10.2 10.2.0 – 10.2.7-h* 10.2.7-h38, or 10.2.18-h12+
Older, unsupported All Upgrade to a supported fixed release
Cloud NGFW, Prisma Access Not affected No action

11.2.16 was published on 15 October, a day after the advisory.

Embedded PAN-OS: Siemens added CVE-2026-0322 to SSA-104023 on 10 November 2026 for the RUGGEDCOM APE1808 (all versions running an affected Palo Alto Virtual NGFW). Siemens says to contact its support for patch information.

Prerequisites: network access to the management web interface (the dedicated MGT port, or any interface with a management profile allowing HTTPS), valid high-privilege admin credentials, and an unpatched build.

Identifying affected systems

On each firewall or Panorama you administer (operational-mode CLI):

# 1) Software version: compare against the table above
show system info | match "sw-version|model|hostname"

# 2) Who can reach the management plane? An empty permitted-ip list means "anyone who can route to MGT"
show config running | match permitted-ip

# 3) Data-plane interfaces with a management profile that allows HTTPS
show config running | match "interface-management-profile|https yes"

# 4) Admin accounts and active sessions: look for anything you don't recognize
show admins all
show admins

For fleets, Panorama’s Panorama > Managed Devices > Summary shows the software version per device. Or use the XML API from your management host:

# Read-only version query via the XML API (key generated for a read-only admin role)
curl -sk "https://fw.example.internal/api/?type=op&cmd=<show><system><info></info></system></show>&key=$PANOS_RO_KEY" \
  | grep -oE "<(hostname|sw-version)>[^<]+"

Then check external exposure. From outside your network, confirm that nothing answers on the management ports (TCP 443, plus 22 for SSH) of your firewalls’ public addresses, using an approved scanning source. An internet-reachable PAN-OS login page is a finding in its own right.

Safe validation in an isolated lab

There’s no public trigger, so validation is about your controls, not the bug:

  1. On a lab VM-Series with an evaluation license, run a pre-fix build, then upgrade it to the fixed hotfix for its line.
  2. Confirm your vulnerability scanner or asset inventory flags the old build and clears the new one.
  3. Apply a permitted-ip list and confirm a host outside it can’t reach the login page.
  4. Log in to the lab firewall as an admin from a host outside your normal jump-host range, make a trivial config commit, and confirm both events reach your SIEM with an alert. That’s the detection you’ll rely on in production.

Detection and threat hunting

Palo Alto doesn’t publish host-level IOCs for this one. Hunt on who touched the management plane, from where, and what changed:

  • Source IOCs from the advisory (threat activity and limited exploit attempts): 66.42.72[.]14, 208.83.239[.]114, 149.248.47[.]234. Search firewall traffic logs, management-plane auth logs and any upstream proxy or NetFlow for these. In the web UI log filter: ( addr.src in 66.42.72.14 ) or ( addr.src in 208.83.239.114 ) or ( addr.src in 149.248.47.234 ).
  • Threat Prevention hits: Threat IDs 510046 and 510047 (Applications and Threats content 9159-10287 or later). Filter the threat log on ( threatid eq 510046 ) or ( threatid eq 510047 ). They only fire if management traffic is decrypted and inspected, so silence doesn’t prove much.
  • Admin authentication. Successful admin logins from unexpected IPs, at unusual times, or to accounts that rarely log in. Also look for failed-then-successful patterns. show log system and the System log in the web UI hold the auth events.
  • Config changes. Unexpected commits, new admin accounts, new or changed management profiles, permitted-ip changes, and new certificate or decryption objects in the Config log (show log config).
  • Tech-support file review. If you suspect compromise, generate a tech-support file and work with Palo Alto support. Root-level code execution can hide from the admin UI, so device logs alone aren’t enough to clear a box.

Correlate admin logins with the IOC IPs first, then widen to “any admin login from outside the jump-host range since 1 October”.

Remediation

  1. Upgrade to the fixed hotfix for your maintenance line, or to 12.2.4 / 12.1.10-h2 / 11.2.16 / 11.1.16-h3 / 10.2.18-h12 or later. The vendor says this is the fix.
  2. Lock down management access. Put the management interface on an isolated management network, set permitted-ip to your jump hosts only, and remove HTTPS from management profiles on internet-facing or untrusted data-plane interfaces. This is Palo Alto’s own best-practice guidance, and it’s what drops the score to 8.5.
  3. Rotate admin credentials and enforce MFA for management logins, since the attack starts with a valid admin account. Prune unused admin accounts.
  4. Panorama first. Patch and lock down Panorama before or alongside the firewalls it manages.
  5. Unsupported releases have no fix. Upgrade to a supported, fixed release.
  6. RUGGEDCOM APE1808: follow Siemens SSA-104023 and contact Siemens support.

Interim mitigations (no workaround exists)

The advisory says plainly: no known workarounds. It lists these partial measures for Threat Prevention customers:

  • Route management traffic through a data-plane interface with a management profile, so it passes through policy.
  • Replace the certificate used for inbound management traffic and decrypt that traffic so the firewall can inspect it.
  • Enable Threat Prevention (Threat IDs 510046 and 510047) on that inbound management traffic.

The vendor notes this coverage is limited and attack variations may get through. Restricting who can reach the management plane at all is the far stronger interim control.

References

Verify every version against the vendor advisory before acting. Published 02 Nov 2026.