🔍 Read the full analysis: Understanding “AI-Assisted” Through The PHP2Shell Server Attack on ThorstenMeyerAI.com
Get hardware and tech essentials delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Dream reported that an attacker exploited PHP2Shell flaws and a previously disclosed Linux kernel bug to gain root access on 24 Palo Alto Networks firewalls, then reach an internal server at a U.S. government-affiliated organization. Dream found AI tools in the attacker’s environment but says it has no direct evidence that AI wrote the exploit or scripts. The report also highlights exposed firewall management pages and a gap between the vendor advisory and the release of fixed builds.
Security company Dream says an attacker compromised 24 Palo Alto Networks firewalls and used one to reach an internal server at a U.S. government-affiliated organization, two days after PHP shipped fixes for vulnerabilities now called PHP2Shell. Dream’s October 26 report, updated October 29, describes a chain of known flaws; it also says investigators found AI tools in the attacker’s environment, but no direct evidence that AI wrote the exploit.
According to Dream, the attacker scanned and filtered internet-accessible firewall management pages, narrowing roughly 19,000 apparent devices to 240 targets. An automated attack on October 23 gained root access on 24 firewalls. Dream says the targets were in the United States, Japan, Taiwan and South Korea. It reported that 22 of the rooted firewalls were left with hidden web shells, while two belonged to the government-affiliated organization involved in the later network intrusion.
Dream says the attacker chained PHP2Shell vulnerabilities CVE-2026-109618 and CVE-2026-109619 with Fragnesia, a Linux kernel privilege-escalation flaw tracked as CVE-2026-46300 and publicly disclosed in May. PHP2Shell provided a way to run code on the firewall’s management web server; Fragnesia was then used to raise privileges to root, according to the report. Palo Alto Networks said its Unit 42 researchers independently verified the attacker’s tools and results.
Dream’s timeline says a second firewall at the government-affiliated organization was compromised on October 24. The attacker then used a tunnel to reach an internal firewall management server and began targeting the organization’s Active Directory, which controls access to accounts and systems. Dream counted 25 compromised servers in total: 24 firewalls and that internal server. The source report does not establish what data, if any, was taken or whether the Active Directory attack succeeded.
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.
- 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
Exposed Management Pages Enabled the Breach
The reported intrusion shows how an internet-facing administration interface can turn a software flaw into a route toward a wider organization. A firewall’s management page is used to configure security rules and devices; if reachable from the public internet, it can give attackers a direct target. Dream says the attacker moved from compromised firewalls to the server managing them, then attempted to reach Active Directory, expanding the potential consequences beyond the edge devices.
The timing also matters. PHP released fixes on October 21, and Palo Alto Networks issued an advisory that day, but Dream says fixed PAN-OS builds for the affected 12.1 and 12.2 branches arrived between October 27 and 29. Until fixes were available, Palo Alto’s stated mitigation was to keep the management interface off the internet. The interval created a period in which exposed systems could remain at risk, particularly if organizations had not restricted access or applied available protections.
The report’s AI angle should be read narrowly. Dream found Claude Code installed and DeepSeek and Kimi configured in a security platform called aiscan. Those findings show AI tools were present in the environment, not that they created or operated the attack. Dream says the available evidence does not show a fully autonomous operation. The clearest practical lesson from the account is about exposure, patching and access controls, rather than proof of a new AI-driven attack method.
Products Worth Considering
LG UltraGear 27G640A-B 27-Inch QHD IPS Gaming Monitor, 300Hz, 1ms, G-SYNC Compatible, FreeSync Premium, DisplayHDR 400
Ledger Flex + Folio Crypto Hardware Wallet
Trezor Safe 5 Crypto Hardware Wallet with Color Touchscreen, Black Graphite
As an affiliate, we earn on qualifying purchases.
Two Flaws, One Attack Chain
The PHP project released updates on October 21, 2026, including versions 8.5.12, 8.4.27, 8.3.36 and 8.2.35, according to the source report. Dream says PAN-OS used its own copy of PHP, identified in its analysis as version 8.0.30. Palo Alto Networks published an advisory on the same date, with fixed builds for the affected software branches becoming available later, from October 27 through 29.
The second flaw in the reported chain, Fragnesia, is a Linux kernel vulnerability disclosed in May 2026 with a public exploit, according to Dream’s account and the cited vulnerability records. Combining a route to execute code with a separate way to gain higher privileges allowed the attacker to move from limited access on the firewall to administrator-level control. Dream says the attack batch ran on October 23, approximately two days after the PHP fixes were released.
Dream says its researchers found artifacts associated with the attacker on October 25 and published their account the next day. On October 26, the U.S. Cybersecurity and Infrastructure Security Agency added both PHP2Shell CVEs to its Known Exploited Vulnerabilities catalog and set a deadline for federal agencies, according to the source material. The report’s dates and technical findings are attributed to Dream unless separately identified as statements from Palo Alto Networks.
““no direct evidence” that AI tools wrote the exploit or scripts”
— Dream
AI Use and Network Impact Remain Unclear
The available account does not show whether AI tools generated any exploit code, selected targets, or helped run the attack. Dream reports finding Claude Code and configurations for DeepSeek and Kimi on the attacker’s servers, but explicitly says it has no direct evidence tying those tools to the exploit or scripts. The extent of any human involvement is also not fully detailed.
It is also unclear whether the attacker succeeded in compromising Active Directory, accessed sensitive information, or caused lasting disruption. Dream describes attacks against the directory system but the supplied report does not specify the outcome. The organization’s identity, the full set of affected owners, and whether all exposed systems have since been secured are not stated in the source material.
Patch Status and Intrusion Findings
Organizations running affected PAN-OS branches should consult Palo Alto Networks’ current security advisory and apply the fixed builds relevant to their installations. The mitigation identified in the source report before those builds were available was to restrict management-page access from the internet. Administrators should also review exposed management services, investigate devices for signs of compromise and check whether firewall credentials or connected management systems may have been affected.
Further reporting may clarify the outcome of the Active Directory attack, the number of affected organizations, and whether investigators can establish a direct role for AI tools. CISA’s addition of the PHP2Shell vulnerabilities to its exploited-vulnerability catalog placed a remediation deadline on federal agencies, according to Dream’s account. The scope of remediation and any confirmed consequences beyond the reported server compromises remain to be established.
Key Questions
What was PHP2Shell?
PHP2Shell is the name used for two PHP vulnerabilities, CVE-2026-109618 and CVE-2026-109619. Dream says the attacker combined them with a Linux kernel flaw to gain control of vulnerable firewall systems.
How many systems were reportedly compromised?
Dream reported root access on 24 Palo Alto Networks firewalls and access to one internal firewall management server, for 25 servers in total. It said the targets were located in four countries.
Did AI create the attack?
The available evidence does not establish that. Dream found AI coding and language-model tools in the attacker’s environment, but said it had no direct evidence that they wrote the exploit or scripts.
What should administrators do?
Administrators should check Palo Alto Networks’ advisory, install the applicable fixed PAN-OS release, and keep management interfaces inaccessible from the public internet unless access is tightly restricted. They should also investigate exposed systems for signs of compromise.
Source: ThorstenMeyerAI.com
