Skip to main content
Vulnerability Database/CVE-2026-97791

CVE-2026-97791: Apache CXF Information Disclosure Vulnerability

CVE-2026-97791 is an information disclosure flaw in Apache CXF where STSTokenValidator improperly validates SAML assertions, allowing forged tokens to bypass security checks. This article covers technical details, affected versions, and patches.

Published:

Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

Sources: National Vulnerability Database, CISA KEV, Exploit-DB

CVE-2026-97791 Overview

Apache CXF contains a race condition in the STSTokenValidator component that handles SAML assertion validation. The validator checks whether a SAML assertion is signed by a trusted certificate before forwarding it to the Security Token Service (STS). The validation result is stored in a single object shared across all concurrent requests. A remote, unauthenticated attacker can submit a forged assertion signed with an untrusted certificate during concurrent processing of legitimate requests. The forged assertion may be accepted as trusted without ever reaching the STS. Only services using STSTokenValidator without alwaysValidateToSts set are affected.

Critical Impact

A remote unauthenticated attacker can bypass SAML token validation through a shared-state race condition, allowing forged assertions to be accepted as trusted and causing availability impact to downstream services.

Affected Products

  • Apache CXF versions prior to 4.2.4
  • Apache CXF versions prior to 4.1.9
  • Apache CXF versions prior to 3.6.13

Discovery Timeline

  • 2026-10-09 - CVE CVE-2026-97791 published to NVD
  • 2026-10-15 - Last updated in NVD database

Technical Details for CVE-2026-97791

Vulnerability Analysis

The vulnerability is a race condition [CWE-362] in Apache CXF's STSTokenValidator. The validator determines whether an incoming SAML assertion is signed by a trusted certificate. If the certificate is trusted, the validator returns success without forwarding the assertion to the Security Token Service.

The root flaw is that this validation outcome is stored in an object shared across all concurrent requests. Under concurrent load, one request can observe the validation state produced by another request. An attacker who submits a forged SAML assertion while legitimate requests are being processed can have the forged assertion marked as trusted.

The impact is scoped to availability. The CVSS vector shows no confidentiality or integrity impact but high availability impact, consistent with downstream services receiving unexpected or malformed security contexts.

Root Cause

The root cause is improper synchronization of shared mutable state during token validation. The validation result object does not isolate per-request state, so concurrent requests can read each other's results. This classic time-of-check to time-of-use pattern allows the trust decision for one assertion to be applied to another.

Attack Vector

Exploitation requires network access to a service that uses STSTokenValidator for SAML token validation and does not have alwaysValidateToSts enabled. The attacker sends a crafted SAML assertion signed with an untrusted certificate while legitimate SAML traffic is in flight. No authentication or user interaction is required. The attack succeeds when the attacker's request races with a legitimate request whose validation result is already cached in the shared object. See the Apache Mailing List Thread and OpenWall OSS-Security Post for upstream discussion.

Detection Methods for CVE-2026-97791

Indicators of Compromise

  • SAML assertions arriving at CXF endpoints signed by certificates not in the configured trust store
  • Spikes in concurrent SAML token validation requests from a single or small set of source addresses
  • Downstream service errors or crashes following acceptance of malformed security contexts
  • Absence of corresponding STS validation log entries for assertions that were accepted as trusted

Detection Strategies

  • Enable verbose logging on STSTokenValidator and correlate assertion issuer and signing certificate against the configured trust anchors
  • Compare application-layer accepted-assertion counts against STS backend validation counts to identify discrepancies
  • Monitor for repeated SAML submissions with varying signer certificates against the same endpoint within short time windows

Monitoring Recommendations

  • Track Apache CXF application logs for SAML validation anomalies and ingest them into a centralized SIEM for correlation
  • Alert on concurrent inbound requests to SOAP endpoints that handle SAML tokens when request volume deviates from baseline
  • Review access logs for repeated token submissions from untrusted or newly observed source IPs

How to Mitigate CVE-2026-97791

Immediate Actions Required

  • Upgrade Apache CXF to version 4.2.4, 4.1.9, or 3.6.13 depending on your current major version
  • Inventory all services using STSTokenValidator and verify whether alwaysValidateToSts is set
  • Enable alwaysValidateToSts as an interim control where configuration changes are faster to deploy than patching
  • Restrict network exposure of affected SAML endpoints to known clients while remediation is in progress

Patch Information

The Apache CXF project has released fixed versions 4.2.4, 4.1.9, and 3.6.13. Upgrade paths and release details are available via the Apache Mailing List Thread and the OpenWall OSS-Security Post.

Workarounds

  • Set alwaysValidateToSts to true so every SAML assertion is validated by the STS regardless of local trust check outcome
  • Replace STSTokenValidator with a custom validator that uses per-request state for the duration of upgrade
  • Place a web application firewall or API gateway in front of CXF endpoints to rate-limit SAML submissions and reduce race window exposure
bash
# Configuration example: force STS-side validation on the CXF STSClient
<jaxws:properties>
    <entry key="ws-security.sts.token.always.validate" value="true"/>
</jaxws:properties>

Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

Sources: National Vulnerability Database, CISA KEV, Exploit-DB

Experience the Most Advanced Cybersecurity Platform

See how the world’s most intelligent, autonomous cybersecurity platform can protect your organization today and into the future.