TRENDING
A black manual telephone switchboard with rows of orange jacks and red patch cords plugged into the lower row, above a rotary dial and a handset
November 16, 2026
How to Route Alerts in Alertmanager So Pages Reach the On-Call and Tickets Reach the Queue, and Test Grouping, Inhibition, and Silences Before an Outage Does
A brass nail with a threaded screw section lying on graph paper, a fastener that is both nail and screw
November 16, 2026
Ai2’s XYBench Turns the XY Problem Into a Benchmark, but the Model That Drafted Every Rubric Also Judges the Answers, the Human Baseline Is Graded on Rubrics Read Off Its Own Answers, and the Emphasis Judge Disagrees With the Annotators on 20 to 26 Percent of Validation Responses
An amber warning beacon on a white surface, a flashing-light stand-in for the FBI’s FLASH alert
November 16, 2026
The FBI’s ARTEX Warning Says a Supply Chain Compromise “Enabled Access” to 2,250 Tenants Where Microsoft Counted Sign-In Attempts and Cites “Over 1,600” Consoles Where Censys Counted 1,465, and a Microsoft Post It Cites Mistypes One Indicator in Its Hunting Query
Two eggs and a colour-changing egg timer in a saucepan of water on an induction hob
November 16, 2026
How to Set gRPC Deadlines and Retries in Python, Pass Them Down a Call Chain, and Stop Retries From Charging a Card Twice
A whole lemon and a halved lemon side by side on a black background
November 16, 2026
IBM’s LLMON Turns Prompt-Injection Defense Into a Token Layout, but the “Nearly Zero” Chart Is One Benchmark, the Paper’s Own Red Team Lets 15% to 20% of Indirect Attacks Through, and the Shipped Template Lets a Document Type the Delimiters
16 Nov 2026
SXZ.io SXZ.io
  • Home
Search the Site
Popular Searches:
Technology Amazon AI
Recent Posts
A manual paper cutter with a wooden grid board, a clear plastic guard and a long cutting arm, a picture of text being trimmed to fit
How to Stop Ollama From Silently Cutting Your Prompt, Find the Real Context Window, and Set num_ctx Without Running Out of Memory
November 16, 2026
A brown llama and her gray cria walking across a salt flat beside a red lake in Bolivia
llama.cpp’s v1.0.0 Promises Major Bumps for API and ABI Breaks, but Its Rule Table Lists Only the C API, libmtmd Shares the Version While Its Header Warns of Breaking Changes, and the Release Page Carries a 7-Byte Pointer to a Nightly Build Instead of Binaries
November 16, 2026
A stack of chocolate chip cookies on a white plate beside a cup of tea
NoisyTCP Can Crash a Linux Server Over the Network, but the Fix Has Reached Only Three of the Seven Kernel Lines Still Being Released, Red Hat and Ubuntu Have No Update, and the Researcher’s Own Timings Run From 5 Seconds to 14 Hours
November 16, 2026
SXZ.io SXZ.io
  • Home

Categories

Articles 302 Posts
News 307 Posts
Learning Hub 274 Posts
Home/News/Check Point Says a Gaia Root Flaw Was Exploited Against “a Couple of Dozen” Customers, Its Researchers Tie It to PHP2Shell and an ESXi Wiper, and the Fix Is Listed Only in Jumbo Takes, Not in the October PHP Hotfix
News

Check Point Says a Gaia Root Flaw Was Exploited Against “a Couple of Dozen” Customers, Its Researchers Tie It to PHP2Shell and an ESXi Wiper, and the Fix Is Listed Only in Jumbo Takes, Not in the October PHP Hotfix

CISA listed CVE-2026-123039 as exploited on November 13 after Check Point reported attacks on a couple of dozen customers; Check Point Research ties the flaw to PHP2Shell and an ESXi wiper, the fix...

November 14, 2026 11 Min Read
29

Check Point says attackers exploited a root-level flaw in Gaia, the operating system behind its firewalls and management servers, against “a couple of dozen customers.” The flaw, CVE-2026-123039, is a local privilege escalation, so it needs code that is already running on the appliance. Check Point Research says that, in the intrusion it investigated, that code came from PHP2Shell, the pair of PHP bugs we covered before the patch and after the early release, used against the Mobile Access portal. From root, the researchers say, the attacker planted webshells and a persistent backdoor, moved over SSH to a VMware ESXi host and ran a wiper. CISA listed the Gaia flaw as exploited on Friday, November 13, with a due date of Monday, November 16.

Table Of Content

  • What Check Point has published
  • The advisory, sk1000362
  • The research report
  • The chain, as Check Point Research describes it
  • Timeline
  • Which fix covers which flaw
  • The CVE record and the advisory do not line up
  • Where the accounts differ
  • Checking a gateway: Check Point’s commands, tested on made-up lines
  • What nobody has said yet
  • Sources and what I checked

Two details are easy to miss. Check Point describes its October 23 PHP hotfix as “the PHP 7.4 published fix,” while the Gaia fix appears only in four Jumbo Hotfix Accumulator Takes, so a gateway that took only the October package has closed the door the attackers used but, going by Check Point’s two advisories, not the flaw behind it. And the advisory lists R81 and R81.10 as affected but names no fix for them, only a mitigation. The CVE record’s version ranges also stop short of the advisory’s first fixed Takes, which leaves a few Takes unaccounted for.

What Check Point has published

The advisory, sk1000362

The advisory was created on Sunday, November 8 and last modified on Thursday, November 12. It rates the flaw High, with a CVSS 3.1 score of 7.8, and explains that the Gaia dynamic CLI is served by a root process, dynamic_server, that listens on a local socket, /tmp/dynamic_socket. The socket is world-writable and the process does not authenticate who connects, so “any authenticated local user can run arbitrary commands as root by writing a single request to the socket.” Check Point adds: “The issue is a local privilege escalation and requires the attacker to already have local code execution on the appliance.” It also says: “Check Point observed an active exploit of this vulnerability against a couple of dozen customers.”

The affected products are Force Firewalls, Management and Log Servers, Multi-Domain Security Management and Multi-Domain Log Server, and Maestro Orchestrator; Spark Firewalls and Smart-1 Cloud are not affected. The versions listed are R82.20, R82.10, R82, R81.20, R81.10 and R81, and the page header marks R81 and R81.10 as end of support (EOS). The mitigation is to stop the daemon with dbset process:dynamic_server followed by dbset :save, which makes the Dynamic CLI commands listed in sk144112 fail until it is switched back on.

The research report

Check Point Research’s report, “PHP2Shell Exploitation on Check Point Gateway,” went up at 13:56 UTC on November 12, and the CVE record was published about six minutes later. It says: “Beginning on 30 October 2026, an attacker executed scanning and exploitation activity against organizations with unpatched Check Point Security Gateways, using Mobile Access portals with PHP2Shell as initial access followed by CVE-2026-123039.” The first wave, on October 30, “targeted Israeli organizations and included destructive activity against some of them.” A broader wave on November 2 reached organizations in several countries and sectors “with no evidence of destructive activity.” The report does not say which fix it means by “approximately one week after a security fix became available,” but Check Point’s PHP2Shell advisory dates its hotfix to October 23, seven days before the first wave (my reading).

The chain, as Check Point Research describes it

Step What the report says Artifacts it names
1. Initial access PHP2Shell (CVE-2026-109618 and CVE-2026-109619) against the Mobile Access portal, with “hundreds of segmentation faults” in the portal’s Apache and PHP workers PHP session files containing the marker upload_progress_pwn under /opt/CPcvpn-R81.20/tmp/
2. Root CVE-2026-123039, through the DynamicCLI interface /tmp/dynamic_socket
3. Webshells PHP webshells in the portal’s templates directory, which the portal uses to build its login pages; a small shell first, then a larger one that can run commands, run PHP and take uploads Files named zz*.tpl; uploads land in /tmp/.up.tpl and are copied to challenge_render.tpl
4. Persistence The TSGA backdoor, started from /tmp/CPconfigdaemon and relaunched as slot_CPconfigdaemon, with an embedded TSProxy library that intercepts networking functions /lib/libcpfwcommon.so, /tmp/.bd.log
5. Lateral movement The open-source gogo scanner on the gateway, then SSH to a VMware ESXi host as the vpxuser service account SSH from the gateway to an ESXi host
6. Destruction The destroy_v6 wiper, which zero-fills files before deleting them unless fast mode is set, and a note that rejects ransom talks DREAM_ANNEXATION.txt

On attribution the report is careful. It says TSGA and TSProxy are “modified variants of PhiliKit,” a framework previously documented in attacks on Ivanti appliances. ESET’s own report calls PhiliKit “a new implant in UNC5221’s SPAWN toolset,” and Check Point Research notes that ESET associated it with the China-nexus group UNC5221. But the researchers list features that do not fit that profile (a destructive operation, a focus on Israel, and a note in Hebrew referring to the Israeli-Palestine conflict) and conclude that “the evidence does not allow us to assess with high confidence the true nature of this operation.” They name three scenarios: “the identified actor ran the entire campaign, acted solely as an initial access broker, or took part in a false-flag operation.” They add: “It is also possible that the destructive activity served as a smoke screen.”

The note “uses the Dream Security brand,” and Dream had recently published its own PHP2Shell research on Palo Alto Networks devices, which we covered. The report says this “timing raises the possibility of retaliation,” but that “the available evidence does not establish a clear motive.” I found no statement from Dream about the note in the sources I read, and Dream’s PHP2Shell post, the one the report links, does not mention Check Point or the note (I searched its text for Check Point, annexation, wiper and Israel).

The report names two source addresses, 130.94.19[.]204 and 130.94.19[.]168. RDAP shows both sit in 130.94.0.0 to 130.94.127.255, a block of 32,768 addresses registered to LIGHT NODE LIMITED, which matches the report’s statement that the attacker’s infrastructure was hosted on LightNode VPS servers. Blocking two addresses inside a block that size will not do much on its own (my reading).

Timeline

Date Event Source
Wed Oct 21 PHP ships its PHP2Shell fixes; Check Point creates its PHP2Shell advisory, sk1000302 BleepingComputer; sk1000302
Fri Oct 23, 18:30 GMT Check Point: “provided a Hotfix for CVE-2026-109618 and CVE-2026-109619” sk1000302
Mon Oct 26 CISA adds both PHP2Shell CVEs to its exploited list (due Oct 29); Check Point reissues the R82, R82.10 and R82.20 hotfix packages at 08:40 GMT for standalone deployments CISA KEV JSON; sk1000302
Wed Oct 28, 11:50 GMT R81.20 hotfix package reissued; users of the Mobile Access Portal must install it on Jumbo Take 166 or higher sk1000302
Thu Oct 29, 11:30 GMT Hotfix released for R81.10 sk1000302
Fri Oct 30 First exploitation wave, against Israeli organizations; destructive activity against some Check Point Research
Mon Nov 2 Broader second wave; no destructive activity seen Check Point Research
Thu Nov 5, 07:30 UTC CVE-2026-123039 reserved CVE record
Sun Nov 8 Advisory sk1000362 created sk1000362
Thu Nov 12 Research report published at 13:56 UTC; CVE record published at 14:02 UTC; both advisories last modified Check Point Research; CVE record; sk pages
Fri Nov 13 CISA lists CVE-2026-123039 (due Monday, Nov 16); CISA’s enrichment marks exploitation active, automatable no, technical impact total CISA KEV JSON, catalog 2026.11.13; CVE record
Sat Nov 14, 04:56 UTC CVE record last updated CVE record

By my count from the CISA catalog, this is Check Point’s fifth entry of 2026, on the fourth listing date: June 8 (CVE-2026-50751), July 22 (CVE-2026-16232), September 22 (CVE-2026-93616 and CVE-2026-85102) and now November 13, spaced 44, 62 and 52 days apart. The catalog’s ransomware-use field for this entry reads Unknown.

Which fix covers which flaw

Two Check Point advisories matter here. sk1000302 covers PHP2Shell and says “Check Point products are not vulnerable to CVE-2026-115560,” the third PHP bug. sk1000362 covers the Gaia flaw. Both list the same four Jumbo Hotfix Accumulator Takes as the fix: R82.20 from Take 14, R82.10 from Take 52, R82 from Take 132 and R81.20 from Take 174. BleepingComputer says the same update “addresses both PHP2Shell and the DynamicCLI privilege escalation flaw used in the attacks.”

Gateway state PHP2Shell (CVE-2026-109618, CVE-2026-109619) Gaia flaw (CVE-2026-123039)
Only the standalone PHP hotfix, first offered October 23 Fixed. Check Point calls it “the PHP 7.4 published fix” and says it “is compatible with all Jumbo Hotfix Accumulator Takes” No fix listed: sk1000362 names only the four Jumbo Takes
Jumbo Take R82.20 14, R82.10 52, R82 132 or R81.20 174, or later Fixed Fixed; the fix “blocks and logs unauthenticated calls to the Dynamic CLI command”
R81.10 (EOS) Hotfix released October 29 Listed as affected; no fix listed; mitigation only
R81 (EOS) Listed as affected; no package in the hotfix table Listed as affected; no fix listed; mitigation only
Spark Firewalls Affected; fixed in R82.00.11 and R81.10.17 builds Not affected

Two more details from sk1000302. It says R82.20 firewalls protect themselves with a built-in IPS signature for PHP remote code execution even when IPS and HTTPS Inspection are off, which would put them in a better position than older releases (my reading). And the R81.20 hotfix needed groundwork: “If you use the Mobile Access Portal, this Hotfix must be installed on top of Jumbo Hotfix Accumulator Take 166 or higher.” Which releases were hit is not stated: the only version string in the research report is the directory /opt/CPcvpn-R81.20/tmp/, which suggests, but does not prove, an R81.20 gateway (my reading). Neither advisory page I read gives release dates for the four Takes.

The CVE record and the advisory do not line up

The CVE record, which Check Point published as the CNA, lists two product names, Quantum Security Gateway and Quantum Security Management, and gives affected versions as “or below” a Take. The advisory’s first fixed Takes sit above those ceilings:

Release Record: affected at or below Advisory: first fixed Takes in neither list
R82.20 Take 10 Take 14 11, 12, 13
R82.10 Take 50 Take 52 51
R82 Take 130 Take 132 131
R81.20 Take 173 Take 174 none
R81.10, R81 not listed no fix listed (advisory lists both as affected) not applicable

Neither document says whether those Takes were affected, or whether they exist. A scanner that reads the record alone would also skip R81 and R81.10, and the record does not say whether its two product names cover the advisory’s list, which includes Maestro Orchestrator and Multi-Domain Security Management. NVD showed the status Undergoing Analysis when I checked at 10:25 UTC on November 14 and lists no product configurations yet, so a scanner that depends on NVD’s version data has nothing to match. The comparison of the record and the advisory is mine, not a Check Point statement.

Where the accounts differ

  • Who ran the campaign. Check Point Research lists three scenarios. BleepingComputer’s second report says the attacks “were conducted by an actor that ran the entire campaign, from initial access to lateral movement and launching PHP webshells,” and then adds that they could not be attributed with confidence.
  • The name of the fix. BleepingComputer says Check Point recommends installing “the November Jumbo Hotfix.” The two Check Point pages I read name four Takes and give no dates for them.
  • How many were hit. The advisory counts “a couple of dozen customers,” a figure BleepingComputer repeats. The research report gives no count of gateways, organizations or countries.
  • Which portals are exposed. The report and BleepingComputer center on the Mobile Access portal. sk1000302 lists six exposed features: Mobile Access Portal, Remote Access VPN with SSL Network Extender, Remote Access VPN with SAML Realm authentication, Identity Awareness Captive Portal, UserCheck Portal and Zero Phishing (the first two do not apply to Spark Firewalls).
  • The wiper’s name. The report’s text says destroy_v6; its indicator table lists the wiper’s hash under destory_v6. Search for both spellings.

Checking a gateway: Check Point’s commands, tested on made-up lines

The advisory gives three checks against /var/log/dynamicCli.log. It says legitimate requests use a Clish wrapper, and that “Any command that does not use this wrapper may indicate that a command was executed as root through the socket and requires investigation.” I have no appliance, so I ran the three commands, as printed, against five made-up lines using GNU grep 3.0 in Git for Windows. Line 3 mirrors the webshell command in the report (printf piped to base64 -d); the rest are invented.

cat > dyn.log <<'EOF'
[2026-11-02 08:00:01] Received = #524288# . /opt/CPshared/5.0/tmp/.CPprofile.sh >> /dev/null; export SP_DISTRIBUTE=0; set -o noglob; show version all 2>&1
[2026-11-02 08:00:09] Received = #return_value# id
[2026-11-02 08:01:00] Received = #524288# printf ZmFrZQ== | base64 -d > /tmp/x.tpl
[2026-11-02 08:02:00] Received = #524288# . /opt/CPshared/5.0/tmp/.CPprofile.sh >> /dev/null; export SP_DISTRIBUTE=0; set -o noglob; id; touch /tmp/marker1 2>&1
[2026-11-02 08:03:00] Received = #524288# . /opt/CPshared/5.0/tmp/.CPprofile.sh >> /dev/null; export SP_DISTRIBUTE=0; set -o noglob; touch /tmp/marker2 2>&1
EOF
# check 1: lines that lack the wrapper
grep "Received =" dyn.log | grep -vF "CPprofile.sh >> /dev/null; export SP_DISTRIBUTE=0; set -o noglob;"
# check 2: the system() sink
grep -F "Received = #return_value#" dyn.log
# check 3: commonly abused commands
grep -EI "Received =.*( id;| whoami| /etc/shadow| ncat| nc -| /dev/tcp| curl | wget | base64| useradd| authorized_keys)" dyn.log
Line Content Check 1 Check 2 Check 3
1 wrapper, then show version all no no no
2 #return_value# id flagged flagged no
3 no wrapper, printf ... | base64 -d > /tmp/x.tpl flagged no flagged
4 wrapper, then id; touch /tmp/marker1 no no flagged
5 wrapper, then touch /tmp/marker2 no no no

Check 1 drops any line that contains the wrapper text anywhere, so a line that carries the wrapper plus extra commands passes it (lines 4 and 5), and line 5 gets past all three. Whether real exploit traffic includes the wrapper is not stated in the advisory or the report; the report’s own webshell command contains base64, which check 3 does match. Two copy-and-paste traps showed up as well. Check 1 is printed with a backslash before the pipe; in bash that adds a stray argument, so grep prints grep: : No such file or directory and puts the file name in front of each hit, though the hits themselves are right. Check 3 ends a line with a backslash, and if the line break is lost when you copy it, grep looks for a file whose name starts with a space, exits with status 2 and prints nothing to review. I did not test what a browser copy produces.

After patching, the advisory says to “confirm the box is actually fixed”: ls -l /tmp/dynamic_socket should show srw-rw---- root config (0660), not srwxrwxrwx (0777). Blocked attempts are logged as Rejected connection from unauthorized uid, and “a non-zero or unexpected UID (for example, 99 = nobody) in these lines indicates that an escalation attempt occurred and was blocked.”

A clean result is not proof of a clean box. The report says TSGA includes “binary firewall-log tampering,” and BleepingComputer lists “unexplained gaps in firewall logs” among the signs of compromise, alongside session files containing upload_progress_pwn, unexpected CPconfigdaemon processes or binaries, and “Unauthorized SSH activity from the gateway to internal VMware ESXi hosts.”

What nobody has said yet

  • How many gateways, organizations and countries were hit; the advisory counts customers and the report gives no count.
  • Which gateway releases were compromised, and whether any gateway on a fixed Take or the October hotfix was.
  • How the attacker obtained the vpxuser credentials for the ESXi host; the report names the account, not the source.
  • What the second wave did after access: BleepingComputer writes that “the follow-on actions the threat actors took remain unclear.”
  • When each Jumbo Take shipped, and whether Takes 11 to 13 of R82.20, 51 of R82.10 and 131 of R82 exist or are affected.
  • Whether Dream Security has a position on the note that carries its name.

Sources and what I checked

Check Point: advisories sk1000362 (Gaia flaw) and sk1000302 (PHP2Shell), both read in a headless browser because the pages need JavaScript, and the research report. Records: the CVE record from CVE Services, CISA’s Known Exploited Vulnerabilities feed (catalog 2026.11.13) and the NVD record. Coverage: BleepingComputer on the root flaw and on the PHP2Shell chain and wiper. Background: ESET’s APT activity report for Q4 2025 and Q1 2026, and Dream’s PHP2Shell post.

My own checks: the weekday of every date, the counts of Check Point entries in the CISA feed, the Take comparison between the CVE record and the advisory, the RDAP lookups for 130.94.19.204 and 130.94.19.168, and the grep test above. I did not touch any Check Point appliance, fetch any malware or run any exploit, and I left out the webshell’s hardcoded key, which the report prints.

Tags:

Check PointCISA KEVCVE-2026-123039PHP2ShellPrivilege EscalationVMware ESXiWiper Malware

Share

Level crossing barriers lowering across a road in front of an approaching white van, with red warning lights lit
Previous Post

How to Fix Random 502 Errors Behind nginx: Reproduce the Keep-Alive Timeout Race With uvicorn, Match the Timeouts, and Check Them in CI

A hand holding a clear glass of water with several fingerprints visible on the glass
Next Post

A Report Turns Model Weights Into a GPU Registry, and It Reproduces in Four Seconds, but the Fingerprint Dates the Starting Weights, Not the Training, and the X Post’s “Conclusive” Covers Six Models Where the Report Claims It for One

No Comment! Be the first one.

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Latest
16 Nov
How to Stop Ollama From Silently Cutting Your Prompt, Find the Real Context Window, and Set num_ctx Without Running Out of Memory
16 Nov
llama.cpp’s v1.0.0 Promises Major Bumps for API and ABI Breaks, but Its Rule Table Lists Only the C API, libmtmd Shares the Version While Its Header Warns of Breaking Changes, and the Release Page Carries a 7-Byte Pointer to a Nightly Build Instead of Binaries
Trending
November 16, 2026
How to Stop Ollama From Silently Cutting Your Prompt, Find the Real Context Window, and Set num_ctx Without Running Out of Memory
November 16, 2026
llama.cpp’s v1.0.0 Promises Major Bumps for API and ABI Breaks, but Its Rule Table Lists Only the C API, libmtmd Shares the Version While Its Header Warns of Breaking Changes, and the Release Page Carries a 7-Byte Pointer to a Nightly Build Instead of Binaries
November 16, 2026
NoisyTCP Can Crash a Linux Server Over the Network, but the Fix Has Reached Only Three of the Seven Kernel Lines Still Being Released, Red Hat and Ubuntu Have No Update, and the Researcher’s Own Timings Run From 5 Seconds to 14 Hours
November 15, 2026
How to Route Alerts in Alertmanager So Pages Reach the On-Call and Tickets Reach the Queue, and Test Grouping, Inhibition, and Silences Before an Outage Does
November 15, 2026
Ai2’s XYBench Turns the XY Problem Into a Benchmark, but the Model That Drafted Every Rubric Also Judges the Answers, the Human Baseline Is Graded on Rubrics Read Off Its Own Answers, and the Emphasis Judge Disagrees With the Annotators on 20 to 26 Percent of Validation Responses
November 15, 2026
The FBI’s ARTEX Warning Says a Supply Chain Compromise “Enabled Access” to 2,250 Tenants Where Microsoft Counted Sign-In Attempts and Cites “Over 1,600” Consoles Where Censys Counted 1,465, and a Microsoft Post It Cites Mistypes One Indicator in Its Hunting Query

Related Posts

Rows of server racks in a data center representing network infrastructure targeted by botnets
News

C0XMO Botnet Shows Why Old Router Firmware Still Matters

June 7, 2026
Close-up of a USB flash drive, representing physical data-theft risk in office security incidents
News

Fake IT Support Is Now Walking Through the Front Door

June 7, 2026
A phone security app on a smartphone resting on a laptop keyboard.
News

Everest Forms Pro Flaw Is Being Exploited to Create Rogue WordPress Admins

June 7, 2026
A phone secured by a padlock, illustrating AI data-leak containment and security controls.
News

OpenAI’s Lockdown Mode Is a Data-Leak Brake, Not a Prompt-Injection Cure

June 8, 2026
SXZ.io SXZ.io
  • [email protected]

Categories

Articles
Learning Hub
News

All Rights Reserved by SXZ.io ©2026