Check Point has released emergency fixes for a previously unknown flaw in its security gateways that attackers combined with the PHP2Shell vulnerabilities to take full control of devices at about two dozen customers, in some cases going on to destroy data, BleepingComputer reported Friday. For organizations that rely on Check Point to guard the edge of their networks, the attacks turn a routine patch into a question of whether intruders are already inside.

What Happened

The new flaw, CVE-2026-123039, is a privilege escalation bug: it lets someone who already has a foothold on a device raise their access to root, the all-powerful administrator account. Check Point's advisory sk1000362, updated Thursday, says a component of its Gaia operating system called DynamicCLI runs as root and carries out commands from any local user without checking who sent them. The company rates the flaw 7.8 out of 10 and wrote that it "observed an active exploit of this vulnerability against a couple of dozen customers."

The advisory covers Force firewalls, management and log servers and Maestro Orchestrator, among others, on six software releases. Hotfixes cover the four supported releases; the two oldest, past end of support, have no listed fix. Spark firewalls and Smart-1 Cloud are not affected.

RelatedCitrix Fixes 10 Critical NetScaler Flaws; No Exploitation Reported So Far

Attackers first broke into the gateways' Mobile Access portal, the web page employees use to connect remotely, through PHP2Shell, a pair of critical flaws in the PHP programming language that were patched Oct. 21, SecurityWeek reported. They then used the new bug to gain root, according to a Check Point Research report published Thursday.

The first wave began Oct. 30, about a week after a fix was available, and targeted Israeli organizations, with destructive activity against some of them, the researchers wrote. The attackers planted web shells, hidden pages that accept remote commands, and two backdoors, then moved to a VMware ESXi server, which runs virtual machines, and ran a wiper that overwrites files with zeros and deletes them. It leaves a politically themed note in Hebrew that uses the brand of security firm Dream Security and rejects any ransom negotiation.

A broader second wave from Nov. 2 reached organizations in multiple countries and sectors, with no destructive activity seen. Check Point found no clear sector preference in either wave, suggesting opportunism.

Why It Matters

A gateway sits between a company's network and the internet, so root control of one gives an intruder a trusted position from which to reach the systems behind it. This is the third exploited Check Point zero-day in about two months; the company warned of two others in September, BleepingComputer noted. Other edge devices have also been PHP2Shell targets: The Primary Wire reported in October that Palo Alto Networks had confirmed attacks on exposed firewall management interfaces, citing the company's advisory.

Ransomware scrambles data and sells the key; a wiper destroys it, so recovery depends on backups the attackers could not reach. Boards should ask whether backups of virtualization servers are isolated.

A hotfix closes the hole but does not remove backdoors already planted. Check Point published indicators of compromise, and administrators should look for root commands that bypassed the gateway's normal command shell, BleepingComputer reported. Customers that cannot install the hotfix right away can stop the vulnerable service and keep it from restarting, the advisory says, which disables some command-line features.

Who Was Behind It

Check Point stopped short of naming a culprit. Its researchers said the backdoors are modified versions of PhiliKit, a toolkit that security firm ESET linked to UNC5221, a China-nexus group, and that scanning ran during Chinese working hours. But the destructive, Israel-focused activity does not fully fit that group's known behavior, they wrote. Dream had recently published research on China-linked PHP2Shell attacks, which Check Point said raises the possibility of retaliation.

"Taken together, the evidence does not allow us to assess with high confidence the true nature of this operation," the researchers wrote. Possibilities include that the actor ran the whole campaign, acted only as a broker of initial access for others or staged a false flag.

What to Watch

As of Friday morning, CVE-2026-123039 was not in the Cybersecurity and Infrastructure Security Agency's Known Exploited Vulnerabilities catalog, whose newest entries were dated Tuesday. A listing would give federal civilian agencies a deadline to patch. The Primary Wire reported in October that CISA had already added both PHP2Shell flaws to the catalog.

Also watch for reports of destructive attacks outside Israel and whether other researchers firm up who was responsible.