TL;DR
- Threat actors began exploiting unpatched Check Point Security Gateways on 30 October 2026, combining PHP2Shell web access with a local privilege escalation in the DynamicCLI IPC interface (CVE-2026-123039) to gain root control.
- Compromised appliances were backdoored using the TSGA malware and TSProxy shared library (/lib/libcpfwcommon.so), derived from the PhiliKit framework, enabling covert traffic interception.
- Attackers scanned internal subnets with the gogo tool and moved laterally to VMware ESXi hypervisors over SSH using stolen vpxuser credentials, executing the destructive destroy_v6 Linux ELF wiper in targeted intrusions.
- CISA added CVE-2026-123039 and CVE-2026-114713 to its KEV Catalog on 13 November 2026, requiring Federal Civilian Executive Branch agencies to remediate exposed assets under BOD 26-04.
- Defenders must immediately audit gateway session directories for upload_progress_pwn markers, verify shared library integrity, restrict ESXi SSH access, and apply emergency vendor patches.
On 30 October 2026, security operations teams observed active exploitation against perimeter Check Point Security Gateways running the Mobile Access portal. Attackers targeted exposed web management interfaces to establish an unprivileged footprint before escalating to administrative control. The threat actors targeted perimeter edge infrastructure by chaining initial web vulnerabilities with cve-2026-123039 exploitation, allowing full root takeover of affected appliances. The initial attack campaign focused on organisations in Israel and included destructive file wiping, before expanding into a broader global wave on 2 November 2026 that prioritised broad access collection without wiper execution.
According to analysis released by Check Point Research, the entry vector relied on PHP2Shell vulnerabilities (associated with CVE-2026-109618 and CVE-2026-109619). Once inside the portal service context, attackers abused an incorrect permission assignment within the gateway’s DynamicCLI Inter-Process Communication (IPC) subsystem (CVE-2026-123039). This privilege escalation enabled the installation of persistent implants, internal network discovery using custom binaries, and lateral movement to hypervisor infrastructure hosting critical enterprise workloads.
In response to active exploitation, the Cybersecurity and Infrastructure Security Agency (CISA) added CVE-2026-123039 and CVE-2026-114713 to its Known Exploited Vulnerabilities (KEV) Catalog on 13 November 2026. Under Binding Operational Directive (BOD) 26-04, Federal Civilian Executive Branch (FCEB) agencies must prioritise remediation of publicly accessible perimeter assets that grant complete system takeover post-exploitation. System administrators and security engineers must evaluate their exposure, inspect appliances for post-compromise indicators, and harden internal hypervisor management channels against SSH-based lateral movement.
Anatomy of an Edge Compromise: Chaining PHP2Shell and CVE-2026-123039
Perimeter network security appliances represent primary targets for operational threat actors because they reside outside traditional endpoint detection and response (EDR) coverage while possessing elevated trust across internal network segments. The campaign targeting Check Point Security Gateways highlights how unauthenticated remote code execution flaws in web presentation layers can be combined with local IPC permission flaws to compromise perimeter defenses completely.
The operational activity unfolded in two distinct operational phases:
- Phase 1 (30 October 2026): Initial exploitation targeted Israeli entities originating from IP address
130.94.19[.]204. This phase combined perimeter compromise with destructive wiper deployment against internal VMware ESXi hosts. - Phase 2 (2 November 2026): A broader scanning and exploitation campaign targeted global organisations across diverse geographies and industry sectors. This second wave reused the initial access chain but omitted destructive payloads, focusing instead on foothold establishment and asset mapping.
The exploitation sequence relies on a multi-stage dependency chain. Unauthenticated web requests trigger PHP2Shell conditions in the Mobile Access portal service. While the portal web server executes under restricted service account privileges, the attackers immediately invoke the local DynamicCLI vulnerability (CVE-2026-123039) to break out of the web container and obtain root access on the underlying Linux operating system.
+-----------------------------------------------------------------------------------+
| 1. Remote Exploitation (PHP2Shell: CVE-2026-109618 / CVE-2026-109619) |
| - Targets Check Point Mobile Access Portal |
| - Obtains unprivileged web server context (Apache/PHP worker) |
+-----------------------------------------------------------------------------------+
|
v
+-----------------------------------------------------------------------------------+
| 2. Local Privilege Escalation (CVE-2026-123039) |
| - Abuses incorrect permission assignment in DynamicCLI IPC interface |
| - Elevates portal service account to full root context |
+-----------------------------------------------------------------------------------+
|
v
+-----------------------------------------------------------------------------------+
| 3. Persistence & Passive Backdooring |
| - Drops zz*.tpl webshells into portal template directories |
| - Deploys TSGA backdoor (/tmp/CPconfigdaemon -> slot_CPconfigdaemon) |
| - Injects TSProxy (/lib/libcpfwcommon.so) into appliance networking daemons |
+-----------------------------------------------------------------------------------+
|
v
+-----------------------------------------------------------------------------------+
| 4. Internal Reconnaissance & Lateral Movement |
| - Executes gogo scanner binary on gateway to map internal subnets |
| - Connects to VMware ESXi hypervisors over SSH using vpxuser credentials |
+-----------------------------------------------------------------------------------+
|
v
+-----------------------------------------------------------------------------------+
| 5. Destructive Action (Wave 1 Intrusion Only) |
| - Deploys destroy_v6 Linux ELF wiper to /tmp/des wiper on ESXi host |
| - Recursively zero-fills files and drops DREAM_ANNEXATION.txt ransom notes |
+-----------------------------------------------------------------------------------+
The inclusion of CVE-2026-123039 in the CISA KEV Catalog under BOD 26-04 mandates that federal agencies conduct aggressive threat-hunting queries before applying updates. Because perimeter appliances lack standard security telemetry, compromise verification requires direct inspection of local file systems, process memory, and temporary execution paths.
Local Privilege Escalation Mechanics in the DynamicCLI IPC Interface
Check Point Security Gateways utilize a specialized command-line interface framework known as DynamicCLI to allow administrative engines, portal components, and system management utilities to query and modify appliance configurations. Communication between unprivileged presentation components (such as the web-based Mobile Access portal) and core system services relies on Inter-Process Communication (IPC) mechanisms, including local Unix domain sockets and IPC message queues.
CVE-2026-123039 stems from an incorrect permission assignment vulnerability within this DynamicCLI IPC interface. Under normal system operations, high-privilege system operations requested via DynamicCLI should require strict privilege verification, validating that the requesting process possesses authenticated administrative capabilities. However, the IPC endpoint failed to properly validate the process credentials or access tokens of client connections, allowing low-privileged execution contexts—such as the Apache web server user—to execute arbitrary administrative commands.
Privilege Escalation Risk
Local privilege escalation flaws on edge security appliances eliminate security boundaries between external-facing web services and internal network controls. Gaining root access allows attackers to disable logging daemons, clear audit trails, dump memory, and manipulate packet filtering rules directly within the operating system kernel.
When the PHP2Shell exploit achieves initial execution within the Mobile Access portal process, the payload interacts directly with the local DynamicCLI IPC socket. By issuing crafted IPC requests across this boundary, the unprivileged web worker instructs the underlying administrative daemon to execute system commands on its behalf. Because the daemon runs with full system authority, the injected commands execute as root.
Gaining root privileges allows the attacker to bypass operating system file integrity controls, install persistent system daemons, modify kernel modules, and establish network hooks that hide operational traffic from local diagnostic tools like netstat or tcpdump. Exact CVSS v3/v4 numerical scores for CVE-2026-123039 have not been published in the source materials, but its presence in CISA KEV reflects its critical role in total system compromise.
Forensic Indicators of Initial Access and Template Webshell Deployment
The initial access phase leaves specific forensic footprints across web server file structures and system log files. The PHP2Shell exploitation technique exploits PHP session file handling during multipart file upload processing. When attackers issue malicious web requests to the Mobile Access portal, temporary session files are written to disk under the portal working directory:
/opt/CPcvpn-R81.20/tmp/
Forensic analysis of compromised gateways reveals session files within this path containing the distinctive upload_progress_pwn session marker. This string represents an artifact generated during the payload staging process. Simultaneously, the repeated memory corruption resulting from exploit attempts triggers hundreds of segmentation faults across Apache and PHP worker processes. System administrators inspecting system log files will observe elevated process crash counts occurring within narrow timeframes.
# Example: Searching for PHP2Shell session markers in portal temporary paths
grep -rnw '/opt/CPcvpn-R81.20/tmp/' -e 'upload_progress_pwn' 2>/dev/null
# Example: Auditing log files for cluster worker segmentation faults
grep -i 'segfault' /var/log/messages /var/log/httpd/* 2>/dev/null | tail -n 20
After achieving privilege escalation to root via CVE-2026-123039, the attackers establish secondary web-based access by staging Base64-encoded PHP webshells. These webshells are placed inside the template directories used by the Mobile Access portal to render dynamic login interfaces. The initial drop commands utilize standard shell pipelines:
printf <base64 Webshell> | base64 -d > <templates>/zz<random>.tpl
The uploaded files adhere to a strict naming pattern: zz*.tpl (for example, zz2026.tpl or randomized variations). The attackers initially deploy a lightweight command-execution shell before replacing it with a fully featured, persistent administrative webshell. The upgraded webshell features the following characteristics:
- Access Control: Requires a hardcoded authentication key (
zz2026) supplied in HTTP requests to unlock command execution functions. - Obfuscation Handling: Accepts both plain-text command parameters and Base64-encoded payload strings processed through XOR obfuscation routines to evade basic web application firewall (WAF) signatures.
- Functional Modes: Supports arbitrary system shell execution, dynamic PHP code evaluation, and filesystem manipulation.
- File Staging: Uploaded files are initially written to a fixed temporary path at
/tmp/.up.tpl. Subsequent HTTP commands instruct the webshell to copy this file into the web server’s production template directory aschallenge_render.tpl.
| Artifact / Indicator | Path / Specification | Forensic Context |
|---|---|---|
| PHP Session Marker | /opt/CPcvpn-R81.20/tmp/sess_* |
Contains string upload_progress_pwn from PHP2Shell exploit. |
| Initial Webshell Template | <templates>/zz*.tpl |
Base64-decoded PHP backdoor embedded in portal layout path. |
| Staged Upload Payload | /tmp/.up.tpl |
Temporary file destination for attacker file uploads. |
| Production Shell Copy | <templates>/challenge_render.tpl |
Renamed persistent webshell endpoint for web access. |
| Web Control Key | Parameter: zz2026 |
Hardcoded authentication key required to execute commands. |
Deep Dive into Persistence: TSGA Backdoor and TSProxy Function Interception
To ensure access that survives web service restarts or web directory remediations, the attackers transition from web-based shells to a persistent, binary backdoor framework. The primary persistent executable, identified as TSGA, is initially staged and executed from temporary operating system storage:
nohup /tmp/CPconfigdaemon >/tmp/.bd.log 2&1 & printf started
When launched with root privileges, the CPconfigdaemon binary carries out a self-installation sequence. It copies itself into the system’s boot persistence configuration directory, renaming its binary to slot_CPconfigdaemon to mimic legitimate Check Point system daemons. Output from the backdoor process is suppressed and directed to a hidden log file located at /tmp/.bd.log.
The core capability of the TSGA framework relies on an embedded shared library payload called TSProxy. During initialization, TSGA extracts this library to disk at the following system location:
/lib/libcpfwcommon.so
Once written to disk, TSGA uses process injection techniques (such as ptrace or dynamic linker preload manipulation) to force libcpfwcommon.so into running appliance networking daemons. Once loaded into process memory spaces, TSProxy intercepts standard C library (libc) and network system calls (such as accept, read, recv, and connect).
+-----------------------------------------------------------------------------------+
| Attacker Network Traffic |
| (Contains magic header bytes / custom handshake) |
+-----------------------------------------------------------------------------------+
|
v
+-----------------------------------------------------------------------------------+
| Network Socket / Network Interface Layer |
+-----------------------------------------------------------------------------------+
|
v
+-----------------------------------------------------------------------------------+
| Target Appliance Daemon (e.g., Apache / CP-Daemon Process Space) |
| |
| +-----------------------------------------------------------------------------+ |
| | TSProxy Hook (/lib/libcpfwcommon.so) | |
| | Intercepts: accept(), recv(), read() | |
| +-----------------------------------------------------------------------------+ |
| | |
| +-----------------+-----------------+ |
| | | |
| v v |
| [ Attacker Traffic? ] [ Normal Traffic? ] |
| | | |
| v v |
| Routes to TSGA Backdoor Routine Passes unhindered to original |
| (Hidden Command Shell) appliance service handler |
+-----------------------------------------------------------------------------------+
This inline hooking mechanism enables covert communication channels:
- Standard incoming network traffic passes through the hooked functions unaltered, preserving normal appliance portal functions.
- When incoming network connections contain specific cryptographic markers or header patterns matching the backdoor protocol, TSProxy intercepts the socket handle.
- The intercepted socket is redirected internally to the TSGA backdoor module, opening an interactive shell session for the attacker without opening new listening ports or triggering firewall alert thresholds.
According to Check Point Research, binary analysis of TSGA and TSProxy demonstrates extensive code reuse, configuration layout structures, and cryptographic routines shared directly with PhiliKit. PhiliKit is a modular malware framework previously documented by security researchers (including ESET) during intrusion campaigns targeting Ivanti network appliances.
Lateral Pivoting to Virtualisation Infrastructure via SSH
After establishing persistent control over the security gateway, the attackers transitioned from initial access to internal network discovery. Perimeter appliances typically maintain interface bindings across multiple internal VLANs and management zones, making them ideal vantage points for internal subnet mapping.
The attackers dropped a compiled scanning binary known as gogo onto the compromised gateway. Executed directly from the gateway’s root shell, gogo performed automated port scanning and service enumeration across accessible internal IP ranges, targeting infrastructure management protocols.
# Generic representation of internal subnet enumeration from compromised gateway
/tmp/gogo -ip 192.168.10.0/24 -p 22,443 -o /tmp/.scan_results.txt
The internal reconnaissance identified hypervisor management interfaces, specifically VMware ESXi hosts. The attackers established outbound SSH connections from the gateway directly to the target ESXi management interfaces. To authenticate, the threat actors used valid administrative credentials for the vpxuser account.
The Role of vpxuser in VMware Environments
The vpxuser account is an automated service account created on individual ESXi hosts when managed by VMware vCenter Server. It possesses full administrative rights on the host to manage virtual machines, datastores, and host configurations. Compounding this risk, flat enterprise network designs frequently allow SSH (port 22) traffic between edge appliances and management interfaces without network-level access controls.
By leveraging valid vpxuser credentials over SSH, the attackers achieved command-line access on core hypervisor hosts without needing to exploit vulnerabilities on the ESXi operating system itself. This highlights the risk of flat infrastructure architecture, where perimeter security gateways share direct, unrestricted management connectivity with critical virtualisation platforms.
Destructive Payload Breakdown: The destroy_v6 Linux ELF Wiper
During the Phase 1 intrusion wave (30 October 2026), attackers deployed a custom Linux ELF executable named destroy_v6 to execute file-wiping operations on compromised ESXi hypervisors. The binary was staged on the hypervisor file system at:
/tmp/des wiper
The destroy_v6 utility is designed for total data destruction across targeted directory trees. It accepts command-line arguments defining target root paths, operational execution flags, and exclusion parameters.
# Example command-line invocation syntax for the destroy_v6 wiper
/tmp/des\ wiper -w "/boot|/lib64|/etc/vmware" /vmfs/volumes/datastore1
Wiper Execution Modes and Options
The wiper incorporates specific command-line switches to modify its operational behavior:
- Default Mode (Zero-Fill Overwrite): The binary recursively traverses the target file system tree. For every file encountered, it opens the file handle and overwrites all existing data blocks with null bytes (
0x00). After the zero-fill pass completes, the wiper calls system unlinking functions to delete the file pointers, preventing standard file recovery or un-erase operations. - Fast Mode (
-f): When the-fflag is passed, the wiper skips the time-consuming zero-fill byte overwrite routine. Instead, it unlinks and deletes files immediately to maximize deletion speed across large virtual disk stores (VMDK files). - Exclusion Filter (
-w): Accepts a pipe-separated string argument specifying directory paths or patterns to exclude from destruction (for example,-w "/sys|/proc"), ensuring the host operating system remains operational long enough to finish destroying non-excluded data paths.
Ransom Note Generation
During data destruction, destroy_v6 drops a politically themed text file named DREAM_ANNEXATION.txt into every traversed directory. According to source details released by Check Point Research, the text inside the note includes precise execution timestamps, incorporates branding referencing “Dream Security”, and explicitly states that no ransom payment negotiations will be accepted.
===================================================================
DREAM_ANNEXATION
===================================================================
Timestamp: 2026-10-30T14:22:01Z
Status: All virtual disks and system datastores zeroed.
Ransom negotiations are REJECTED. Data recovery is IMPOSSIBLE.
Branding: Dream Security
===================================================================
Campaign Attribution, False Flags, and Geopolitical Context
Attributing edge appliance intrusion campaigns presents complex forensic challenges, particularly when threat actors reuse shared malware frameworks, public vulnerability disclosures, and false-flag artifacts.
The technical connection between TSGA/TSProxy and the PhiliKit framework links this campaign to prior threat intelligence disclosures. ESET previously documented PhiliKit as a signature toolset utilized by China-nexus threat actor UNC5221 during high-profile zero-day exploitation of Ivanti devices. However, security analysts at Check Point Research emphasize that technical evidence alone remains inconclusive regarding definitive threat actor attribution.
Analysts highlight several plausible operational scenarios:
- Single State-Sponsored Actor: Threat actor UNC5221 executed the complete campaign, transitioning from Ivanti targets to Check Point perimeter devices using custom PHP2Shell chains and local privilege escalation.
- Initial Access Broker (IAB) Model: An independent initial access broker executed the automated perimeter scanning and exploitation (deploying webshells via CVE-2026-123039), subsequently selling access to separate operational groups with differing motives.
- False-Flag Operation: A distinct threat actor integrated publicly available PhiliKit code samples and configuration patterns to frame China-nexus groups while pursuing independent operational objectives.
- Smoke-Screen Destructive Activity: The destructive wiping observed in Phase 1 against Israeli targets served as a high-visibility smoke screen designed to mask secondary cyber espionage, persistent backdoor installation, or data exfiltration across broader network environments.
The ransom note dropped by the wiper references “Dream Security”. Dream Security is a cybersecurity research firm that had recently published technical analysis on PHP2Shell exploitation affecting Palo Alto Networks firewalls by China-nexus groups. While the inclusion of this brand name suggests potential retaliation or intentional misdirection, available evidence cannot establish a definitive motive or confirm actor identity.
Furthermore, the contrast between Phase 1 (30 October 2026, targeted destructive wiping in Israel) and Phase 2 (2 November 2026, non-destructive global access collection) suggests that the attackers shifted operational focus from targeted destruction to opportunistic access aggregation across vulnerable edge infrastructure worldwide.
Detection, Forensics, and Mitigation Guidance for Defenders
Securing perimeter infrastructure against chained vulnerabilities requires immediate patch management, process integrity auditing, and network segmentation. Because edge appliances sit outside traditional endpoint security tools, incident response teams must use explicit diagnostic commands and targeted log analysis to verify system integrity.
Mandatory Remediation Priority
Under CISA BOD 26-04, agencies and enterprise operators must remediate KEV-listed vulnerabilities on perimeter assets immediately. Prior to applying updates, teams must execute forensic triage commands to verify that threat actors did not establish persistence prior to patch installation.
1. Immediate Patch Application
System administrators must apply official security updates for Check Point Security Gateways addressing CVE-2026-123039 and associated Mobile Access portal vulnerabilities (CVE-2026-109618, CVE-2026-109619). Specific software version patch build numbers containing official vendor remedies have not been detailed in public source disclosures; administrators must consult official vendor support portals to download verified update packages.
2. Appliance Integrity Triage Script
Execute the following diagnostic shell commands via SSH as root on Check Point Security Gateways to search for forensic artifacts associated with this campaign:
#!/bin/bash
# Check Point Gateway Forensic Triage Script (Test in Staging First)
echo "[+] Checking for PHP2Shell session indicators..."
grep -rnw '/opt/CPcvpn-R81.20/tmp/' -e 'upload_progress_pwn' 2>/dev/null
echo "[+] Inspecting portal template directory for unexpected zz*.tpl webshells..."
find /opt/CPcvpn-R81.20/ -name "zz*.tpl" -o -name "challenge_render.tpl" 2>/dev/null
echo "[+] Checking staging file locations..."
ls -la /tmp/.up.tpl /tmp/.bd.log /tmp/CPconfigdaemon 2>/dev/null
echo "[+] Checking for TSProxy shared library installation..."
ls -la /lib/libcpfwcommon.so 2>/dev/null
echo "[+] Auditing running processes for slot_CPconfigdaemon persistence..."
ps aux | grep -i 'slot_CPconfigdaemon' | grep -v grep
echo "[+] Auditing Apache and PHP crash events..."
dmesg | grep -i 'segfault' | tail -n 15
3. Hypervisor Isolation and Access Hardening
Because attackers leveraged SSH to move laterally from compromised gateways to ESXi hypervisors, platform architects must enforce strict administrative access controls:
- Disable ESXi SSH Services: Disable SSH services on VMware ESXi hosts by default. Enable SSH only temporarily during active troubleshooting windows.
- Restrict Management Access: Implement firewall rules and dedicated management VLANs to ensure hypervisor management interfaces (port 22 and port 443) cannot be reached directly from edge security gateways or general employee subnets.
- Audit vpxuser Activity: Audit ESXi authentication logs located at
/var/log/auth.logor/var/log/shell.logfor abnormalvpxuserSSH interactive logons originating from non-vCenter IP addresses.
Common Operational Mistakes
| Mistake | Operational Consequence | Corrective Action |
|---|---|---|
| Patching without forensic triage | Overwrites volatile evidence while leaving persistent backdoors (e.g. TSProxy hooks) running in process memory. |
Perform memory and file system integrity checks before initiating appliance reboot or upgrade cycles. |
| Relying solely on external port scans | Fails to detect passive backdoors (TSProxy) that reuse existing web ports without listening on new ports. |
Inspect local process memory maps and audit file paths like /lib/libcpfwcommon.so directly via shell access. |
| Flat network connectivity to hypervisors | Allows compromised perimeter appliances to act as direct entry points for internal hypervisor takeover. | Enforce micro-segmentation, isolating ESXi management interfaces behind dedicated administrative jump boxes. |
| Leaving ESXi SSH services running | Provides direct SSH attack surfaces for automated lateral movement via stolen vpxuser service credentials. |
Disable ESXi SSH services persistently across all hosts in vCenter cluster policy settings. |
Frequently asked questions
What is CVE-2026-123039 and why is it critical for Check Point gateway administrators?
CVE-2026-123039 is an incorrect permission assignment vulnerability in the DynamicCLI IPC interface of Check Point Security Gateways. It allows local, unprivileged service accounts—such as Apache web portal workers—to execute administrative system commands as root. Its presence in CISA’s KEV Catalog reflects its active use in perimeter compromises to grant attackers total appliance control.
How do attackers move from web portal access to root privilege on the Check Point appliance?
Attackers gain initial access by exploiting PHP2Shell flaws (CVE-2026-109618 / CVE-2026-109619) in the Mobile Access portal, obtaining an unprivileged web process shell. They then send malicious Inter-Process Communication (IPC) messages to the DynamicCLI interface. Because DynamicCLI fails to validate requesting process permissions, it executes the payload as root, granting administrative privileges.
What forensic artifacts indicate a successful PHP2Shell and TSGA compromise?
Key artifacts include PHP session files containing upload_progress_pwn in /opt/CPcvpn-R81.20/tmp/, high rates of Apache/PHP worker segfaults, webshells matching zz*.tpl or challenge_render.tpl in template directories, execution logs at /tmp/.bd.log, persistent processes named slot_CPconfigdaemon, and the presence of /lib/libcpfwcommon.so.
How does the TSProxy shared library hide attacker network traffic on the appliance?
TSProxy (/lib/libcpfwcommon.so) extracts from the TSGA backdoor and injects into running appliance process memory spaces. It hooks standard networking system calls like accept and recv. TSProxy inspects incoming network traffic: standard web connections pass to normal services, but connections with specific attacker markers are silently intercepted and routed to the backdoor shell.
Why were VMware ESXi hosts targeted following the gateway compromise?
Compromising perimeter gateways gives threat actors access to internal subnets. Attackers ran the gogo scanner on the gateway to map internal management networks and identified VMware ESXi hypervisors. By extracting or abusing legitimate vpxuser service account credentials over SSH, attackers gained direct administrative execution capabilities on ESXi hosts without needing additional hypervisor exploits.
Sources and further reading
- CISA Adds Two Known Exploited Vulnerabilities to Catalog | CISA (CISA Advisories)
- PHP2Shell Exploitation on Check Point Gateway – Check Point Research (Check Point Research)
- PHP2Shell Exploited in Destructive Check Point Gateway Attacks (SecurityWeek)
This article was drafted with AI assistance from the publicly cited sources and passed automated checks for structure and sourcing. Verify critical details against the vendor advisories linked above before acting.