CVE-2026-90436: end-of-life Apache Axis 1.x added to CISA KEV — unauthenticated class instantiation through SOAP faults

CISA added the EOL Apache Axis 1.x remote-class-instantiation flaw to its Known Exploited Vulnerabilities catalog on Nov 18, 2026, with a federal remediation deadline of Nov 21. There is no patched release; operators must migrate or rebuild.
CoeLab Real-World Impact: 94/100 — Urgent. Editorial score v1.0. Evidence checked 2026-11-19T13:38:00.158937+00:00. CVSS and EPSS are separate measures, not this editorial score.
Rating breakdown: exploitation: 30, reachability: 25, consequences: 24, operational relevance: 15.
Who is affected
CVE-2026-90436 affects every release of Apache Axis 1.x, up to and including 1.4, the final release of a Java SOAP stack Apache retired long ago. The CVE record tags it "unsupported when assigned," and the affected Maven coordinates include both the legacy axis:axis artifact and org.apache.axis:axis, spanning essentially all published versions that applications might still be pulling from Maven Central.
Because Axis 1.x is a library rather than a stand-alone product, exposure is indirect. Any Java web application that still runs on Axis 1.x and accepts inbound SOAP requests — custom enterprise services, older internal integrations, and a long tail of systems whose SOAP endpoints were set up in the 2000s and never revisited — is a candidate. The upstream record identifies Maven Central as the distribution point, so dependency trees built years ago can still resolve it today.
A version check on its own is not enough to prove exploitability. Exploitation requires a suitable class on the application's classpath, as described below, so two applications on the same vulnerable Axis build can differ sharply in real risk. Discovery of the flaw is credited to José Apari-Pantigozo of OKIOK.
Real-world impact and exploitation evidence
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 with a base score of 9.8 (critical): network-reachable, low complexity, no privileges or user interaction, and high scores across confidentiality, integrity and availability. The CISA coordinator's SSVC assessment, timestamped 2026-11-17, marks Exploitation as active, Automatable as yes, and Technical Impact as total.
The strongest new evidence is the KEV entry: CISA added CVE-2026-90436 to the Known Exploited Vulnerabilities catalog on November 18, 2026, listed as "Apache Axis Unsafe Reflection Vulnerability," with a required-action due date of November 21, 2026, and marked for forensic triage under BOD 26-04. Ransomware campaign use is recorded as Unknown. It is important to be precise about what this means: the vulnerability itself is not new. The CVE was published on November 6, 2026, and the KEV addition twelve days later is the news — CISA is reporting actual exploitation of a flaw in abandoned software, which is consistent with attackers favoring targets that will never receive an official patch.
Evidence about exploitation is nonetheless thinner than the KEV headline suggests. EPSS on 2026-11-18 was about 1.1 percent — low, and typical of a niche component where scanning targets are scarce. CISA's KEV entry does not name observed threat actors, campaigns or victim sectors, and the project's own advisory provides no attack telemetry. So the honest reading is: credible exploitation evidence (KEV inclusion with an active-exploitation assessment), but unquantified breadth, and a low probability estimate for any single random internet host being attacked today.
Where exploitation occurs, the impact ceiling is high. If an attacker-chosen class can be loaded or instantiated inside the application, consequences can escalate to arbitrary code execution in the context of the SOAP engine's process, depending on the class available.
Attack prerequisites
The flaw is CWE-470, unsafe reflection driven by externally controlled input. Axis 1.x parses the fault section of inbound SOAP responses, and if a <Fault> carries a <detail><exceptionName> element, Axis loads the class named there. If a companion <data> element supplies a string, the class is instantiated through its String constructor — before Axis checks that the class is actually an AxisFault. Even without a <data> element, loading the class alone executes its static initializer.
Three prerequisites therefore shape real-world exploitability: (1) the attacker's input must reach Axis's fault-processing code path, which per the vendor advisory happens as the engine resolves the target service — importantly, before any per-service handler chain runs, so authentication handlers configured on a service do not protect against it; (2) a suitable class must be present on the application's classpath, i.e., a class whose static initializer or String constructor performs something attacker-valuable when the string is attacker-controlled; and (3) the attacker must be able to submit the crafted SOAP fault to the service. The classpath condition is the great attenuator: common JDK classes historically used in such chains exist in most JVMs, but whether a given application's path leads to code execution is an application-specific question we recommend auditing rather than assuming either way.
Authentication posture at the application level deserves skepticism here. Because the fault is processed while the engine resolves the target service, prior to the handler chain, wrapping a vulnerable Axis service in per-service authentication does not close the hole. Network-level exposure is what matters: a system reachable only from a trusted internal segment presents a materially smaller attack surface than one exposed to the internet, though CISA's KEV listing shows that wherever vulnerable systems are reachable, someone is using it.
Actions and fixes
There is no fixed Axis 1.x release and, per the Apache advisory, none is planned — Axis 1 is end-of-life and its final release, 1.4, remains affected. The vendor's recommendation is to migrate to a different SOAP engine, such as Apache Axis 2/Java. Migration is the only durable fix, and the KEV due date of November 21, 2026 sets a federal deadline within three days of this post.
As a stopgap for teams that cannot migrate immediately, Apache points to a patch commit in the axis-axis1-java repository, which changes class loading so that classes named in untrusted input are located without being initialized — avoiding the static-initializer execution that turns class loading into code execution. Applying that change requires building your own copy of Axis from source with the patch applied; it is not available as an official release artifact, and a self-built fork carries its own maintenance obligations you must own.
Operational triage should follow the KEV entry's own logic: first inventory whether Axis 1.x is present anywhere in dependency trees (checking both coordinate forms in Maven Central naming); then identify which of those services accept SOAP requests and from where; treat internet-exposed instances as the urgent tier and apply BOD 26-04-style prioritization even if you are not a federal agency. Wherever a vulnerable service exists but no longer serves a purpose, decommissioning it is simpler and more secure than patching. Until a service is fixed or removed, tightening network access to the SOAP endpoint is a real, if partial, mitigation — it does not repair the flaw, but it removes the reachability that exploitation requires.
Because the record is flagged for forensic triage, organizations with Axis 1.x deployments should check available logs for suspicious SOAP fault traffic before remediating, since exploitation leaves an artifact trail in the request handling path that a standard access log may not capture.
Uncertainty and what could change
The biggest open question is the scope of exploitation behind CISA's decision to add this to KEV. The catalog entry affirms exploitation is occurring, and CISA's coordinator assessment marked the vulnerability active and automatable, but neither source quantifies who is exploiting it, against what targets, or at what volume. Ransomware association is formally Unknown. If CISA subsequently updates the entry with named threat actors or campaign detail, the practical outlook — especially for internal-only deployments — should be revisited.
There is also genuine uncertainty about exploitability per deployment. The advisory describes the mechanism but does not enumerate which classpath-resident gadgets lead from class loading or String-constructor instantiation to meaningful impact; that determination depends on each application's dependency set. Operators should avoid the two extremes of panic (assuming every Axis 1.x install is trivially compromised) and complacency (assuming the classpath condition makes it unreachable).
A versioning nuance worth noting: the CVE record lists the Maven artifact axis:axis as defaultStatus affected, while org.apache.axis:axis and the git source range are marked with defaultStatus unaffected at the record level, yet both carry the same affected version range in their entries. The vendor description is unambiguous that all Axis 1.x versions through 1.4 are affected, so we treat the full range as affected and flag the record's field-level inconsistency rather than inventing a distinction between Maven artifact generations. The record was last updated on 2026-11-19, after the KEV addition, so downstream feeds may continue to normalize this over the coming days.
Sources and evidence date
This analysis draws on four primary sources, each verified directly: the Apache Software Foundation's vendor advisory and the CVE record assigned by Apache's own CVE Numbering Authority — the authoritative statements on affected versions, mechanism and remediation; the corrective patch commit in the apache/axis-axis1-java repository, which documents the removed code path and confirms the static-initializer risk in the maintainer's own comments; and CISA's Known Exploited Vulnerabilities catalog entry and the CISA-ADP enrichment data in the CVE record, which supply the exploitation assessment and deadline.
On dating, the evidence timeline is: discovery credited externally, CVE published by Apache on 2026-11-06 (with an oss-security reference in the CVE program container); CISA's SSVC enrichment timestamped 2026-11-17; KEV addition on 2026-11-18 with a due date of 2026-11-21; the CVE record updated 2026-11-19 to reflect the KEV status. The Apache advisory page itself is also referenced by the record and used to source the migration-versus-rebuild guidance. This article reflects the state of those sources as of 2026-11-19; if CISA or Apache publish further updates, the numbers and threat picture above may change and should be rechecked against the linked records.
Scoring note: this story's CoeLab Real-World Impact v1.0 score of 94 (Urgent) is an editorial screening tool from our published methodology, not CVSS or EPSS. Its components here are Exploitation 30/30 (KEV listing is weighted at the top of that scale), Reachability 25/25 (network vector, no privilege, low complexity), Consequences 24/25 (9.8 CVSS base), and Operational Relevance 15/20 (recognized platform-class library rather than edge or core infrastructure). Two limitations matter for interpreting it: the score treats KEV presence as full exploitation evidence even though the catalog entry does not detail the campaigns, and its operational-relevance field is a category heuristic that cannot see whether your organization actually still runs Axis 1.x — the 94 reflects the published picture, not your network. That is precisely why we publish the analysis rather than only the number.
Primary-source links
https://lists.apache.org/thread.html/8slf0j9gz35z56t0glt5h0ncwlbn4bdz
https://github.com/apache/axis-axis1-java/commit/41f3b4e8d04e310ca81f86b92630ab2c11d608bd
https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2026-90436
https://www.cve.org/CVERecord?id=CVE-2026-90436
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.
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.
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.