CVE-2026-123039: Check Point Local Socket Flaw

CISA added CVE-2026-123039 to its exploited-vulnerabilities catalog on 13 Nov 2026. Patch your Check Point gateways by 16 November — here's how.

Maysan Engineering Team · · 5 min read

Xin

On 13 November 2026, CISA added CVE-2026-123039 to its Known Exploited Vulnerabilities catalog. The entry covers Check Point Firewall, Multi-Domain Security Management Server, Scalable Platforms, Security Gateway and Security Management Server, and it describes an incorrect permission assignment for a critical resource: the Gaia OS DynamicCLI daemon runs as root and listens on a world-writable local socket without authenticating the processes that connect to it, so an authenticated local user can send a command to that socket and have it executed as root. The required action is to apply mitigations in line with vendor instructions by 16 November 2026. If any of those products are in your estate, this is a this-week job, not a this-quarter one.

What the flaw actually is: root, a world-writable socket, no authentication

This is a local privilege-escalation problem, not a remotely reachable network exploit, and that distinction should drive your sequencing. The DynamicCLI daemon runs as root. Its local socket is world-writable. It does not authenticate the processes that connect to it. The result is that anyone who already holds an authenticated local account on the box can send a command to that socket and get it executed with root privileges. The catalog entry does not describe which commands, what preconditions apply beyond local authentication, or any exploitation detail beyond that mechanism — so plan your exposure assessment around "who can log in locally to this device" rather than around internet-facing attack paths alone.

Who CVE-2026-123039 affects: Check Point Multiple Products

Check Point Multiple Products is the affected family named in the catalog, and CISA lists five product lines inside it:

  • Check Point Firewall
  • Multi-Domain Security Management Server
  • Scalable Platforms
  • Security Gateway
  • Security Management Server

If any of those sit in your estate, you are in scope — including standby cluster members and lab or DR copies sharing the same build. CISA's required action also addresses cloud services: follow applicable BOD 26-04 guidance for cloud services, or discontinue use of the product if mitigations are unavailable. The entry does not name a fixed version, so the vendor's own instructions are the only authoritative source for what "mitigated" means on your build. Known ransomware campaign use is listed as unknown, which is an absence of reporting, not a clean bill of health.

Your next 24 hours: apply the mitigation, then prove what is serving traffic

Apply Check Point's mitigations exactly as the vendor instructs, then verify the running build on the device actually handling traffic instead of trusting the management console's inventory. In practice:

  • Pull the mitigation or update from the vendor channel, not a mirror or a colleague's copy.
  • Apply it per cluster member, fail over, and confirm the passive node carries the same state.
  • On Multi-Domain Security Management, check each domain, not only global management.
  • Record the build string you observed on each device, with a timestamp and who checked it.
  • Flag any device you could not verify as unresolved, not as done.

The verification step is where teams lose time. The console shows what was pushed, not necessarily what is running now; reading the build back from the live device closes that gap.

What to watch in the logs afterwards

Treat the first 48 hours after patching as a monitoring window, because a fixed socket does not tell you whether it was used before the fix. Watch for unusual local logins and for processes spawned under the DynamicCLI daemon or inside its process tree; depending on how your Gaia logging is configured, exported audit and system logs are where this surfaces. Root-owned child processes that don't map to a known administrative action deserve a second look, as do changes to local user accounts and any command execution that traces back to the socket rather than to an interactive admin session. Retain those logs past your normal rotation window — if the forensics triage material CISA points to becomes relevant, you will want raw records, not a summary.

What AI-assisted review changes about a bug like this

Nothing public says how this particular flaw was found, so be sceptical of anyone who tells you otherwise — but the workflow around this bug class has genuinely shifted. The pattern here is a privileged daemon, a world-writable socket, and no authentication of the connecting peer, and that combination is now a searchable anti-pattern. AI-assisted review is good at the boring part: sweeping large diffs and IPC-heavy code for sockets, pipes and local endpoints where the connecting process is trusted by assumption. It is weak at the part that decides your risk: proving who can actually reach that endpoint in your deployment. Use it to shorten the list of places a human must reason about, then keep the reasoning human. On the response side, the same tooling is useful for mapping a catalog entry against your asset inventory and finding the DR box nobody remembered.

The 16 November deadline, and what BOD 26-04 asks of you

The required action is not "schedule a patch"; it is to apply mitigations in accordance with vendor instructions, ensuring compliance with CISA's BOD 26-04 prioritisation guidance and its Forensics Triage Requirements — or to discontinue use of the product where mitigations are unavailable. CISA also puts the exposure question on you: stakeholders are responsible for evaluating each asset's internet exposure and for adhering to BOD 26-04 patching guidelines. That matters even for a local bug. Management interfaces reachable from the internet are how an attacker gets the local account this vulnerability needs in the first place, and the architecture decisions about which management planes are exposed are the control that outlives the next CVE too.

Where to start

Pull your Check Point asset list today, including standby members and DR, and establish which of them can authenticate a local user who should not have root. If that list lives in a spreadsheet maintained by hand, fixing that is worth more than this single patch: we have built inventory and patching dashboards for teams in exactly this position, and you can see how we approach cloud and infrastructure work before deciding anything. Maysan Labs builds this kind of internal tooling — the point is that when the next due date lands on a weekend, you already know which boxes are affected.

Buying software or a website?

Twelve questions to ask any vendor before you sign: price, your data, fit for an Indian business and support.

Get the free checklist

Written by Maysan Engineering Team at Maysan Labs, a software studio in Gurgaon, India. Follow us on LinkedIn