Palo Alto Networks confirmed on October 28, 2026, that attackers are exploiting the Palo Alto PHP2Shell flaws against PAN-OS management interfaces reachable from the internet. If a firewall sits in front of your website, its admin page may now be the weakest door into your business.
Key takeaways
- Palo Alto Networks confirmed active PHP2Shell exploitation of internet-exposed PAN-OS management interfaces on October 28, 2026.
- CVE-2026-109618 and CVE-2026-109619 (CVSS 9.2) affect PAN-OS 12.2, 12.1, 11.2, 11.1 and 10.2; fixes come only through PAN-OS releases.
- CISA set an October 31 deadline for Fragnesia (CVE-2026-46300), which chains with PHP2Shell to run commands as root.
- Dream reported 25 rooted servers and follow-on Active Directory attacks but did not name the firewall vendor.
- Restrict management access to trusted internal IPs now; it reduces risk but does not replace the patch.
What did Palo Alto Networks confirm on October 28?
Palo Alto Networks confirmed that attackers are actively exploiting the PHP2Shell flaws against PAN-OS management interfaces exposed to the internet. The company added the warning to its advisory on Wednesday, October 28, 2026: “Active exploitation has been observed targeting the Palo Alto Networks management interface exposed to the internet.”
PHP2Shell is the name for two critical flaws in the PHP interpreter itself, CVE-2026-109618 and CVE-2026-109619. Each scores 9.2 on CVSS and lets an attacker with no account run code on a system using an affected PHP build. PAN-OS ships PHP inside its management web interface, so the firewall inherits the problem.
Two details in the advisory explain the urgency. First, no special PAN-OS configuration is needed for a device to be affected. A firewall on an affected release, configured normally, is in scope.
Second, risk peaks when the management interface accepts connections from external internet addresses. If you can load your firewall’s login page from hotel Wi-Fi, assume someone is probing it this week. Attackers do not wait for your maintenance window, and scanners will find an exposed appliance within days.
Why doesn't updating WordPress or your site's PHP close this?
Because the vulnerable PHP here runs inside the firewall, not on your web server. Updating WordPress, Joomla or the PHP version in your hosting panel protects the site. It does nothing for the PHP build bundled into PAN-OS.
The PHP Foundation shipped upstream fixes on October 21, 2026, in PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35. Your web servers still need those updates. On a Palo Alto device, though, you cannot swap out the interpreter yourself. The fix arrives only as a PAN-OS maintenance release from the vendor.
Palo Alto drew a tight boundary around the exposure: “This issue strictly affects the management web interface, which includes the HTTPS web UI and XML API service.” Visitor traffic passing through the firewall to your site is not the attack path.
A dataplane interface, one of the ports carrying ordinary traffic, is in scope only if someone attached a management profile to it. That happens more often than teams admit, usually to make remote admin easier during a migration, and then nobody removes it.
Panorama, the console that manages fleets of Palo Alto firewalls, appears in the same access-restriction guidance. The reporting does not include a separate Panorama affected-version list, so check the advisory directly if you run it.
Why should a website owner care about a firewall bug at all? An attacker who controls the box in front of your site sits on the path to everything behind it: admin logins, VPN sessions, internal servers and the directory that ties them together. Dream’s report, covered below, shows exactly that kind of follow-on.
Which PAN-OS versions are affected by the Palo Alto PHP2Shell flaws?
Five PAN-OS branches are affected: 12.2, 12.1, 11.2, 11.1 and 10.2. The advisory splits them into 20 maintenance-line fix thresholds, so the fixed build depends on the maintenance line you run, not just the branch number.
Several fixes were still pending when the warning went out. These are the ETAs listed in the advisory:
| PAN-OS fix build | Listed ETA |
|---|---|
| 12.1.4-h13 | October 29, 2026 |
| 10.2.10-h42 | October 29, 2026 |
| 12.1.10-h3 | October 30, 2026 |
| 11.2.4-h23 | October 30, 2026 |
| 11.1.7-h12 | October 30, 2026 |
| 10.2.7-h39 | October 30, 2026 |
Those dates come from the October 28 advisory update. Before you schedule a change window, confirm in the advisory that your build has actually shipped; an ETA is not a download.
For every other line, take the exact threshold from Palo Alto’s advisory. We are not reproducing the full matrix, because a wrong digit in a hotfix suffix is how teams end up patched on paper and exposed in production.
To see what you run, use show system info on the CLI and read the sw-version field, or check the General Information widget on the dashboard. Compare the full string, including the -h hotfix suffix, against the row for your maintenance line.
What can you do if your PAN-OS fix is not out yet?
Restrict who can reach the management interface, today. Palo Alto’s own guidance: “The security risk posed by this issue on PAN-OS and Panorama is minimized by restricting access to the management interface to only trusted internal IP addresses.” In practice that means a short permitted-IP list, a dedicated management network, and no internet exposure whatsoever.
For releases without a fix, the vendor lists interim protections:
- Route MGT-port traffic through a dataplane (DP) port, so management sessions pass through the firewall’s own inspection engine.
- Replace the Certificate for Inbound Traffic Management.
- Decrypt inbound management traffic so it can be inspected.
- Enable threat prevention on that traffic.
The first three steps exist to make the fourth work. Threat prevention cannot block an exploit it cannot see inside encrypted traffic.
Some customers got coverage early. Those with Frontier Virtual Patching (ATP Plus) on PAN-OS 12.2.4 and later received Threat ID 543004 at 10:21 PM PT on October 20, 2026, the day before PHP published its fixes. If you hold that subscription, confirm the signature is present and set to block, not just alert.
An IP allowlist cuts reachability; it does not remove the bug. A compromised workstation, a stolen VPN account or a jump server can still reach an internal-only interface. Treat the restriction as a stopgap until the patch lands, not a replacement for it.
How does the Fragnesia chain turn a web shell into root?
Fragnesia gives an attacker who already runs code on a Linux system a way to become root, and PHP2Shell supplies that first foothold. Tracked as CVE-2026-46300, Fragnesia is a local privilege-escalation flaw in the Linux kernel with a CVSS score of 7.8 (High).
The chain has two steps. PHP2Shell lets an unauthenticated attacker run code through the web interface, typically by planting a PHP web shell: a small server-side script that takes commands through web requests. That shell starts with the limited rights of the web server process.
Fragnesia removes those limits. CISA put it plainly: “This vulnerability can be chained with CVE-2026-109618 and CVE-2026-109619 to implant PHP web shells and run commands as root.”
CISA added CVE-2026-46300 to its Known Exploited Vulnerabilities catalog with an October 31, 2026 deadline. That date binds U.S. federal civilian agencies; for everyone else, a KEV entry is a strong signal to move the item up the queue. Note that this entry is for Fragnesia, the kernel bug, not for the two PHP flaws.
No fixed kernel versions were named in the reporting, and it does not spell out how Fragnesia applies to PAN-OS itself. On your own Linux servers, watch your distribution’s security notices for CVE-2026-46300.
Root changes the recovery math. With root, an attacker can rewrite files, create accounts, tamper with security tools and edit local logs. Deleting the one web shell you found proves nothing about what else was left behind.
Can the Dream report be pinned on Palo Alto?
Not yet. Dream, an Israeli security company, described the campaign without naming the firewall vendor: “The affected devices run specific versions of a large vendor’s firewall software.”
According to Dream, a suspected China-nexus actor chained PHP2Shell with Fragnesia and rooted 25 servers in the United States, Japan, Taiwan and South Korea. Three sat in a U.S. government-affiliated environment that Dream did not name.
Those three are 12% of the total, and they do not mean three federal agencies were breached. The reporting does not say who operated those systems.
The intrusions did not stop at the edge. “The attacker then launched attacks against Active Directory. The recovered artifacts confirm credential extraction and directory enumeration.” That matters even if none of the rooted machines were domain controllers. Credentials pulled from one host let the actor log in elsewhere as a legitimate user, which ordinary alerting rarely flags.
The Hacker News suggested a possible Palo Alto link, based on the advisory timing and the absence of exploitation reports from other network-security vendors. That is inference, not confirmation. The outlet said it had contacted both Dream and Palo Alto Networks and had not received answers when it published.
Remember too that 25 is a confirmed incident count from one firm, not an estimate of global exposure. The pattern is familiar: edge appliances keep becoming long-lived footholds. We saw a similar playbook in the F5 BIG-IP PoisonedRefresh campaign, where attackers hid a web shell in appliance memory.
What should site owners and admins check today?
First, find out whether you have a Palo Alto device and who manages it. Many small businesses run a firewall installed by an MSP or hosting provider and have never logged into it. Send your provider these questions today:
- Which PAN-OS version and hotfix is running, and is it at or above the advisory threshold for that maintenance line?
- Is the management web UI or XML API reachable from the internet? The permitted IP list lives under Device, Setup, Interfaces, Management.
- Does any dataplane interface have a management profile attached?
- If the fix is still pending, which interim protections are in place?
- Any new admin accounts, unfamiliar files or unexpected outbound connections from the device since October 20?
Practitioner’s note. In our engagements, the exposed management page is rarely the one the team knows about. It is the second interface opened during a migration, the XML API kept alive for an old automation script, or a management profile left on an outside-facing port.
Test from outside: from a network you do not control, try HTTPS on every public IP the firewall answers on. Any login page is your first fix. Attackers find these pages the same way they find weak websites, by scanning public address ranges, which we covered in how attackers find exposed admin surfaces.
If compromise is plausible, do not trust host logs alone. Pull firewall traffic, VPN, identity provider, EDR and Active Directory authentication records from systems the attacker could not edit. Look for logins originating from the firewall’s own address or its service accounts.
If you find signs of compromise, preserve evidence, rebuild from known-good images instead of cleaning in place, and rotate every credential stored on or used through the device: admin accounts, VPN users, LDAP bind accounts and API keys. For IOCs, use the vendor advisory and the researchers’ published reports.
Then check the sites behind it. A rooted firewall that decrypts traffic can capture WordPress or Joomla admin logins in transit, so reset those passwords and scan the sites for changes; SiteGuarding Antivirus can confirm whether files were altered.
While you are there, verify the PHP your own stack actually uses. Containers, control panels and PHP-FPM pools can run separate builds, so one php -v result does not prove every service is on a fixed release.
What is still unknown?
Several gaps remain, and they affect how you scope your response:
- Which vendor’s firewalls Dream saw rooted. The Palo Alto connection is inferred, not stated.
- The exact date Dream observed the campaign, beyond days before October 29.
- Which Panorama versions are affected, if any, beyond the access guidance.
- Fixed Linux kernel versions for Fragnesia, and how it applies to PAN-OS.
- How many Palo Alto management interfaces are reachable from the internet worldwide.
- Any attribution beyond Dream’s “suspected China-nexus” label.
Until those answers arrive, treat any affected PAN-OS device with an internet-facing management page as a target, and check it as one.
FAQ
Is Palo Alto Networks being exploited with PHP2Shell?
Yes. Palo Alto Networks confirmed on October 28, 2026, that attackers are exploiting CVE-2026-109618 and CVE-2026-109619 against PAN-OS management interfaces exposed to the internet.
Does updating PHP on my website fix the Palo Alto flaw?
No. The vulnerable PHP runs inside PAN-OS, so only a PAN-OS maintenance release fixes the firewall. Your web servers still need PHP 8.5.12, 8.4.27, 8.3.36 or 8.2.35 separately.
How do I check if my Palo Alto management interface is exposed?
From a network you do not control, try reaching each public IP of the firewall over HTTPS. Also review the permitted IP list on the management interface and look for management profiles on dataplane interfaces.
What is Fragnesia CVE-2026-46300?
It is a Linux kernel local privilege-escalation flaw scored CVSS 7.8. CISA says it can be chained with the PHP2Shell flaws to plant PHP web shells and run commands as root.
What should I do if a PAN-OS patch is not out yet?
Limit management access to trusted internal IPs, then apply Palo Alto's interim steps: route MGT traffic through a DP port, replace the inbound management certificate, decrypt that traffic and enable threat prevention.
If your firewall or provider can’t answer the questions above, start with the sites you control. A SiteGuarding website protection package adds monitoring, a firewall layer and hack recovery if something already got through.
