The Coe Lab
← Back to Security

CVE-2026-121653: Pre-auth SQL injection to remote code execution in Ivanti Endpoint Manager Core, fixed in EPM 2024 SU8

By November 15, 202612 min read
CVE AnalysisCVE-2026-121653Cybersecurity
A technology notebook with visual symbols for AI, cybersecurity, infrastructure, and automation

Ivanti patched an unauthenticated SQL injection in EPM's Core services that leads to remote code execution. If you run EPM, the practical takeaway is simple: get to 2024 SU8.

CoeLab Real-World Impact: 87/100 — Urgent. Editorial score v1.0. Evidence checked 2026-11-15T10:10:40.289130+00:00. CVSS and EPSS are separate measures, not this editorial score.

Rating breakdown: exploitation: 18, reachability: 25, consequences: 24, operational relevance: 20.

Who is affected

CVE-2026-121653 is a SQL injection vulnerability in the Core components of Ivanti Endpoint Manager (EPM). Ivanti, the CVE Numbering Authority for this record, lists all EPM releases before version 2024 SU8 as affected, and 2024 SU8 as the unaffected (fixed) version. If a deployment is on Service Update 8 or later of EPM 2024, the vendor's record marks it clear. Anything older needs updating.

Endpoint Manager is Ivanti's on-premises device-management platform: a central Core server that communicates with agents across Windows workstations and servers, handles software distribution, patch deployment and inventory. That architecture matters for exposure math. In most EPM deployments the Core server is reachable from the managed fleet, and in a meaningful fraction of deployments it is also reachable from less trustworthy network segments — a management VLAN, a DMZ, or in some cases the internet for remote offices.

Only the Core is implicated by the vendor's description. Managed endpoints running EPM agents are not described as independently vulnerable; the agent population's risk flows through the compromise of the server that commands them. Ivanti published the advisory on 2026-11-10 and updated a day later; both dates fall a week behind this analysis, so any environment still unpatched has now had the better part of a week to act.

One boundary worth stating plainly: the vendor record uses "before version 2024 SU8" as the affected range and does not enumerate per-branch build numbers. Whether EPM 2025 or later branches exist and are covered is not specified in the record we reviewed. Organizations running a version newer than 2024 should verify their build against Ivanti's advisory before assuming they are unaffected.

Real-world impact and exploitation evidence

The failure is textbook: improper neutralization of input destined for a SQL query (CWE-89), reachable by a remote, unauthenticated attacker, and — per the vendor's own description — sufficient for remote code execution. In an endpoint-management product, the server that executes the vulnerability is also the server that can push software and scripts to every managed machine. RCE on the Core therefore is not a single-host compromise in practice; it is a natural pivot point to fleet-wide code execution, credential caches, and the database EPM maintains about every managed device.

That said, the exploitation picture as of this writing shows no confirmed in-the-wild abuse. CISA's automated vulnrichment (the ADP record attached to the CVE entry, timestamped 2026-11-11) records Exploitation: none, alongside Automatable: yes and Technical Impact: total. The CVE is not on CISA's Known Exploited Vulnerabilities catalog as of 2026-11-15. EPSS, the community exploitability forecast from FIRST, sits at roughly 1.98% (81st percentile) as of 2026-11-14 — elevated for a week-old CVE but far from the levels seen when a vulnerability has genuinely caught on.

This combination — automatable per CISA, no observed exploitation yet — describes an attack that is plausible and likely unattended-weaponizable but not currently happening at scale. In this domain the pattern is familiar: pre-auth RCE-class flaws in management or edge infrastructure attract mass scanning weeks after disclosure, not days. The honest reading is that the next few weeks carry more risk than the last week did, and the cheapest defense is to close the window before that shift.

It is equally important not to inflate this. Nothing in the public evidence indicates ransomware use, a KEV listing, or public exploit code verified against this specific vulnerability. Claims that EPM deployments are 'under attack' would not be supported by the sources reviewed here.

Attack prerequisites

The published CVSS 3.1 vector is CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H. Reading the prerequisites out of it: the attacker needs network access to a vulnerable EPM Core service (Attack Vector: Network), no special privileges or credentials (Privileges Required: None), no user interaction, and no conditions that make exploitation timing- or environment-dependent (Attack Complexity: Low).

In other words, the only prerequisite is reachability. An attacker who can open a network connection to the vulnerable Core interface can attempt the injection directly. That distinction — network-accessible versus internet-exposed — is the entire practical risk spectrum for this CVE. An EPM Core reachable only from an internal management segment requires an attacker to first foothold that segment; an EPM Core reachable from untrusted networks or the internet requires nothing of the sort.

Because CISA marks the condition automatable, defenders should assume tooling could be built to scan and exploit it without human involvement. What the public record does not include is the specific vulnerable endpoint or parameter — the vendor's description names the Core generally, and no independent technical detail beyond that has been verified for this analysis.

Scope Unchanged in the vector means the injection is evaluated as a compromise of the Core system itself; the fleet-wide consequences are a deployment reality, not a scored property. Treat them as the reason urgency is high, while remembering the formal severity describes the one system.

Actions and fixes

The fix is version-based: upgrade to Ivanti Endpoint Manager 2024 SU8, the version the vendor's record designates as unaffected. No interim condition that closes the vulnerability is described in the record we reviewed — no configuration toggle, no URL-block rule, no partial mitigation. That absence matters for planning: this is a patch-or-expose decision, not a mitigate-while-you-test one.

A sensible sequence for EPM operators: first, identify where the Core server (or servers, in larger estates) sits on your network and whether it is reachable from untrusted segments — exposure triage can legitimately reorder patching across sites. Second, schedule the 2024 SU8 upgrade, including the staging EPM deployments normally require; Core upgrades in management products are the kind of change that deserves the platform's own validated upgrade path rather than a rushed in-place patch. Third, while the upgrade is pending on any system, tighten network access around the Core: restrict what can reach its service ports to the management segments and agents that genuinely need it.

After patching, review EPM's own logs and the Core's database access history for anomalies going back to at least 2026-11-10, the disclosure date — and ideally further back, since the window during which the flaw existed before disclosure is unknown. Prior Ivanti incident history is a general reason for care with this vendor's edge-and-management products, but we have no evidence that pre-dates this specific CVE's disclosure, and none is claimed here.

The vendor advisory itself is served through Ivanti's support hub, which requires an interactive session to read in some regions; if the link below does not render for you, the same content is reachable through the Ivanti support portal by the article title shown in the source list.

Uncertainty and what could change

Several things remain genuinely unknown rather than merely unconfirmed. The discovery route is recorded as UNKNOWN — we cannot tell whether this was found internally, via bug bounty, or by a third party, and that colors how likely independent rediscovery is. The exact vulnerable interface inside the Core is unnamed in the public record, limiting defenders' ability to write targeted detection rules until someone documents it.

The fixed-version boundary may broaden. The record currently states one affected line (before EPM 2024 SU8); if Ivanti extends guidance to other product branches or issues additional service updates, the effective affected population changes and this analysis would need revision. Watch the Ivanti hub page, which carries the vendor's canonical advisory and is the one source likely to be updated beyond the CVE record.

Exploitation status is the most likely near-term change. Neither a KEV listing nor an EPSS surge has occurred as of 2026-11-15; either event would materially raise urgency without changing the fix. Our data was collected 2026-11-15 from feeds dated 2026-11-14, so there is an inherent one-to-two-day lag; a KEV add could exist by the time you read this that was not yet in the feed. If exploitation does begin in the wild, the assessment shifts from 'patch this week' to 'patch urgently with exposure triage first' — but as of this evidence, the calmer framing is the correct one.

Finally, the CoeLab Real-World Impact score for this CVE is 87 (Urgent). That is our v1.0 heuristic, and it is neither CVSS nor EPSS. Its components here: 18 of 30 for exploitation, from an EPSS of ~1.98% falling in the >=0.01 tier and below the >=0.1 tier, with no KEV listing; 25 of 25 for reachability, from the fully network-exploitable vector (AV:N, PR:N, AC:L); 24 of 25 for consequences, from a 9.8 CVSS base score; and 20 of 20 for operational relevance, from the endpoint-management platform category. Four things a reader should keep in mind: the category match is a heuristic, not a measured installed-base count; EPSS is a 30-day forecast probability, not evidence of actual attacks; the score deliberately omits exposure information it cannot observe, so a firewalled Core scores the same as an internet-facing one; and a 87 is a prioritization signal — the score does not represent any specific environment's actual exposure, including ours.

Sources and evidence date

Primary evidence consists of the CVE record as published by Ivanti in its role as CNA, retrieved 2026-11-15 as revision e5ecea1b from the official CVE List repository, and CISA's automated ADP enrichment (SSVC decision timestamped 2026-11-11) attached to that same record. The Ivanti hub advisory is cited as the vendor's canonical advisory; it is presented through a JavaScript application that did not render for automated retrieval during this analysis, so its content could not be independently confirmed beyond what Ivanti committed to the CVE record itself. That is an accepted limitation noted above.

Timeline, with each event distinguished: CVE ID reserved 2026-11-03; advisory and CVE record published 2026-11-10; record and CISA enrichment updated 2026-11-11; EPSS observed 2026-11-14; no KEV addition as of the 2026-11-15 feed. Affected and fixed versions are as the vendor published them in the CVE record: all EPM before 2024 SU8 affected, 2024 SU8 unaffected.

The CoeLab Real-World Impact v1.0 score (87, Urgent) was computed by the CoeLab screening pipeline from the sources above; its scoring methodology is described at the blog's /blog/security/methodology page. No exploit testing, proof-of-concept verification, or deployment enumeration was performed for this article, and no claim is made about any specific environment's exposure.

Primary-source links

https://hub.ivanti.com/s/article/Security-Advisory-Ivanti-Endpoint-Manager-EPM-November-2026?language=en_US

https://www.cve.org/CVERecord?id=CVE-2026-121653

https://raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/2026/121xxx/CVE-2026-121653.json

AI-assisted research and drafting, with a separate model editorial review. CoeLab ratings are editorial judgments, not a substitute for checking your own versions and vendor advisories. Methodology: https://thecoelab.com/blog/security/methodology

Related Posts

CVE-2026-123890: VMware vCenter authentication bypass exploited since September — CISA patch deadline is November 23

A critical unauthenticated vCenter Directory Service flaw was used in real intrusions weeks before disclosure. CISA gives federal agencies until November 23 — everyone else should not wait longer.

Nov 22, 2026• 12 min

Your USB-C Dock Is Part of Your Network. Test It That Way.

A trending Ethernet investigation shows why docks belong in your network diagram. Here is how to isolate bottlenecks before replacing routers or adapters.

Nov 22, 2026• 9 min

CVE-2026-129448: Apache Roller XML-RPC deserialization gets CISA KEV deadline — unauthenticated remote code execution in versions before 6.1.5

Apache Roller before 6.1.5 runs unauthenticated XML-RPC deserialization before authentication; it is now on CISA's exploited list with a patch deadline of November 23, 2026.

Nov 21, 2026• 11 min