What “AI-Assisted” Really Adds To The PHP2Shell Story
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: What “AI-Assisted” Really Adds To The PHP2Shell Story on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on office and shipping supplies

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

TL;DR

Dream says an attacker used PHP2Shell and a separate Linux kernel flaw to gain root access on 24 Palo Alto Networks firewalls and compromise a 25th server. AI tools were found on the attacker’s infrastructure, but Dream said it has no direct evidence they generated the exploit or scripts. Palo Alto’s fixes for the affected PAN-OS branches arrived October 27–29, after the attacks.

An attacker used PHP2Shell flaws and a previously disclosed Linux kernel vulnerability to gain root access on 24 Palo Alto Networks firewalls and then compromise a 25th server inside a U.S. government-affiliated organization, according to cybersecurity firm Dream. Dream found AI tools on the attacker’s infrastructure, but said it had no direct evidence that AI wrote the exploit or attack scripts—an important distinction as the incident is described as “AI-assisted.”

Dream’s account says the attacker ran an automated batch against 240 firewalls on the evening of October 23, about two days after the PHP project released fixes for two flaws now called PHP2Shell. The attacker obtained root access on 24 devices. The targets were in the United States, Japan, Taiwan and South Korea. Palo Alto Networks said its Unit 42 researchers independently verified the attacker’s tools and results.

Dream says the attacker combined the two PHP vulnerabilities, CVE-2026-109618 and CVE-2026-109619, with Fragnesia, a Linux kernel privilege-escalation flaw assigned CVE-2026-46300 and made public in May. In Dream’s account, PHP2Shell allowed commands to run under a restricted account on the firewall; Fragnesia was then used to obtain administrator-level access. Dream said the full chain, including establishing a persistent back door, took less than 15 minutes.

On October 24, the attacker used two compromised firewalls belonging to the same government-affiliated organization to reach its internal firewall-management server, which Dream counted as the 25th compromised server. The report says the attacker then targeted the organization’s Active Directory, a system used to manage user accounts and access. Dream reported that 22 of the 24 rooted firewalls were left with hidden web shells. It did not establish in the supplied account what data, if any, was taken from the organization.

Dream also found Claude Code installed on the attacker’s servers, along with the DeepSeek and Kimi models configured in an offensive-security platform called aiscan. Those findings show that AI tools were present in the attacker’s environment. They do not, on their own, establish how the tools were used or whether they contributed to the exploit.

At a glance
reportWhen: Attacks reported October 23–24, 2026; D…
The developmentA Dream investigation published October 26 and updated October 29 linked PHP2Shell flaws to a campaign that compromised 25 servers, while reporting AI tools on the attacker’s infrastructure without proving they built the attack.
Thorsten Meyer AI
Timeline

Infographic 01 · PHP2Shell on Palo Alto firewalls

Two days from patch to root.

PHP fixed the bug on 21 October. Fixed firewall software for the attacked branches arrived almost a week later. The attack ran in between.

~2 daysfrom PHP’s fix (21 Oct) to the first rooted firewall (23 Oct), my count from the dates
24 +1firewalls rooted, plus one internal management server (Dream)
<15 minfor the whole chain, up to a root back door (Dream)
22rooted firewalls left “parked” with a hidden web shell (Dream)
  • 13–23 MayFragnesia goes public, kernel fix followsLinux privilege-escalation bug with a public exploit; CVE-2026-46300.
  • 8 OctAttacker sets up an AI platformaiscan configured with DeepSeek and Kimi, per Dream’s timeline.
  • 21 OctPHP fixes PHP2ShellPHP 8.5.12, 8.4.27, 8.3.36, 8.2.35. Palo Alto advisory the same day; PAN-OS fixes “upcoming”.
  • 23 Oct, evening UTCBatch run: 240 targets, 24 rootedFound via device search engine, filtered to US, Japan, Taiwan, South Korea.
  • 24 OctInto a government-affiliated networkTunnel, persistence, internal management server rooted (no. 25), Active Directory attacked.
  • 26 OctDream publishes · CISA lists PHP2ShellFederal deadline 29 Oct.
  • 27–29 OctFixed PAN-OS builds for 12.2 and 12.1CISA lists Fragnesia on 28 Oct; Unit 42 confirms Dream’s results on 29 Oct.

The fix for PHP existed. The fix for the firewall did not, yet. The open management page was the only door that mattered.Thorsten Meyer · my reading

Sources: Dream Research, 26 Oct 2026 (updated 29 Oct); php.net, 21 Oct 2026; Palo Alto Networks advisory (updated 29 Oct 2026); CISA KEV catalog, version 29 Oct 2026; oss-security, 13 May 2026; Unit 42, 29 Oct 2026. Day count: Thorsten Meyer.
Powered by Thorsten Meyer AI

The Patch Gap Left Firewalls Exposed

The incident highlights a gap between releasing a software fix and making that fix available in a product that embeds the software. Dream said PAN-OS used its own copy of PHP, version 8.0.30, and that Palo Alto published a security advisory on October 21 while fixes for the affected 12.1 and 12.2 branches were still forthcoming. According to the source account, fixed builds arrived between October 27 and 29, after the October 23–24 compromises.

For organizations, the practical exposure was not limited to whether they had updated PHP elsewhere: they needed a vendor update for the firewall software itself. Palo Alto’s advisory recommended keeping the management interface off the public internet, a measure the source describes as the available protection while corrected PAN-OS builds were pending. The reported path into an internal management server also shows why a perimeter device’s compromise can matter beyond that device: access may provide a route to other administrative systems.

The AI label matters because it can blur what the evidence actually establishes. Dream’s findings support the narrower statement that AI tools were present on infrastructure linked to the attacker. They do not establish that AI discovered the vulnerabilities, wrote working exploit code, selected victims, or independently carried out the intrusion. The reported operation involved automated scanning and exploitation, but Dream said it did not appear to be fully autonomous. For defenders, the demonstrated weaknesses remain familiar: exposed management pages, an unpatched kernel flaw, and access controls that may not limit follow-on activity.

Roborock F25 GT Set Cordless Wet Dry Vacuum Cleaner and Mop with 2 Brushes

Roborock F25 GT Set Cordless Wet Dry Vacuum Cleaner and Mop with 2 Brushes

Breville Bambino Espresso Machine, Brushed Stainless Steel (BES450BSS)

Breville Bambino Espresso Machine, Brushed Stainless Steel (BES450BSS)

As an affiliate, we earn on qualifying purchases.

From May Kernel Bug to October Attacks

The attack relied on vulnerabilities with different disclosure timelines. Fragnesia was disclosed in May 2026 with a public exploit, according to the source material. The PHP project released updates on October 21 for the PHP2Shell flaws, and Palo Alto Networks published an advisory that day. The reported campaign began on October 23, before fixes for the affected PAN-OS branches were available.

Dream’s account says the attacker first identified roughly 19,000 PAN-OS firewalls with internet-facing management pages using FOFA, a search engine for internet-connected devices. A custom scanner obtained version information from about 5,000; filtering by country and software version reduced the set to 240 targets. These are figures reported by Dream, not a count of devices confirmed vulnerable worldwide.

Dream published its findings on October 26 and updated them on October 29. The source says the U.S. Cybersecurity and Infrastructure Security Agency added both PHP2Shell CVEs to its Known Exploited Vulnerabilities catalog on October 26 and set a three-day deadline for federal agencies. The source also reports that Palo Alto’s fixed builds for the affected branches shipped from October 27 through October 29.

““no direct evidence” that AI tools wrote the exploit or scripts”

— Dream

What the AI Evidence Cannot Show

The available account does not show whether Claude Code, DeepSeek or Kimi contributed to vulnerability research, code development, target selection, or any stage of the intrusion. Dream explicitly said it had no direct evidence that AI tools wrote the exploit or scripts. Their presence is an indicator about the attacker’s working environment, not proof of a causal role.

Other details also remain unresolved in the source material. It does not specify what information the attacker accessed or took after reaching Active Directory, whether every planted web shell remained active, or the full remediation status of all affected systems. The figure of 25 refers to the servers Dream says were compromised in this campaign; it should not be read as the total number of vulnerable firewalls or affected organizations.

Patch PAN-OS and Review Access

Organizations using affected PAN-OS branches should check Palo Alto Networks’ security advisory for the fixed version that applies to their systems and install it, while following the vendor’s current guidance. The source account says management interfaces should not be exposed directly to the internet; administrators should verify that access is restricted to trusted networks and review firewall logs and systems for signs of compromise.

Because the reported intrusion reached an internal firewall-management server and targeted Active Directory, affected organizations may also need to examine administrative accounts, remote access paths, and persistence mechanisms—not just update the perimeter appliances. The source does not provide a public timeline for a further Dream investigation or a complete incident response account from the government-affiliated organization. More detail on the scope of access and any data impact remains pending.

Key Questions

What does PHP2Shell refer to in this report?

It is the name used for two PHP flaws, CVE-2026-109618 and CVE-2026-109619, that Dream says were combined with the Fragnesia Linux kernel flaw in attacks on PAN-OS firewalls.

Did AI write the exploit?

That has not been established. Dream found AI tools on the attacker’s servers but said it had no direct evidence they wrote the exploit or scripts.

How many servers did Dream say were compromised?

Dream reported root access on 24 firewalls and access to one internal firewall-management server, for 25 servers in its account. That is not a count of all potentially vulnerable devices.

When were fixes available for the affected firewalls?

The source material says Palo Alto Networks’ fixed builds for the affected PAN-OS 12.1 and 12.2 branches shipped between October 27 and 29, 2026. Administrators should consult the vendor’s advisory for the appropriate update.

What should firewall administrators do?

Apply the vendor’s relevant fixed PAN-OS release, restrict management interfaces to trusted networks, and review logs and systems for signs of compromise. Organizations that may have been affected should also investigate internal management systems and accounts.

Source: ThorstenMeyerAI.com

This content is for general information only and is not financial, tax or legal advice. Consult a qualified professional for decisions about your money.
You May Also Like

The UK Is A Cautionary Tale For The U.S.

The UK’s recent economic and political struggles highlight risks for the US as it considers similar policy paths, experts warn.

The Regulatory Vacuum.

Google disclosed an AI-discovered zero-day on May 11, 2026, highlighting a lack of regulatory frameworks to manage AI-driven vulnerabilities and exploits.

A Guide To Telegram’s Android Flaw And Possible Next Steps

NebuSec says Telegram fixed an Android call flaw in version 12.10.4. Here is what is known, what remains unverified and what users can do.

The PHP2Shell Story: 22 Years Of Context And AI-Assisted Review

PHP fixed two critical remote code execution flaws on Oct. 21, 2026. Researchers connect the bugs to a longer history of memory-limit issues and AI-assisted review.