Automated Threat Modeling

Brief Summary: IBM API Connect CVE-2026-17131, an Unauthenticated SSRF Outbound Filter Bypass (CVSS 9.3)

A brief summary of CVE-2026-17131, a server-side request forgery flaw in IBM API Connect V12 OnPrem 12.1.0.0 through 12.1.1.2 that lets a remote attacker bypass outbound request restrictions. Covers affected versions and IBM's patch information (upgrade to 12.1.1.3).

CVE Analysis

7 min read

Gaurav Sarraf
Gaurav Sarraf

2026-11-06

Brief Summary: IBM API Connect CVE-2026-17131, an Unauthenticated SSRF Outbound Filter Bypass (CVSS 9.3)
Experimental AI-Generated Content

This CVE analysis is an experimental publication that is completely AI-generated. The content may contain errors or inaccuracies and is subject to change as more information becomes available. We are continuously refining our process.

If you have feedback, questions, or notice any errors, please reach out to us.

[email protected]

Introduction

An API gateway that can be talked into fetching arbitrary internal URLs and handing back the response is effectively an unauthenticated proxy into the network it sits in. That is the practical shape of CVE-2026-17131 in IBM API Connect. A remote attacker with no credentials can bypass an existing outbound request filter, and IBM scores the flaw 9.3 under CVSS 3.1.

IBM API Connect is IBM's API management platform, which IBM describes as a scalable, security-rich solution that brings APIs together in one place. IBM reports that it was named a Leader in the 2026 Gartner Magic Quadrant for API Management for the 11th time, so API Connect sits at the center of many enterprise API programs. Because the platform sits between consumers and backend services, a flaw in how it makes its own outbound requests matters to everything reachable from where it runs.

Technical information

Root cause

The vulnerability is classified as CWE-918, Server Side Request Forgery. MITRE's general definition covers a server retrieving a URL without sufficiently ensuring the request goes to the expected destination.

API Connect already had a control in place for this. IBM's description states that "restrictions on outbound requests can be bypassed". The product enforced an outbound destination filter, and that filter could be evaded. Once it is evaded, the server contacts a destination it should have refused and returns the response to the caller. The fix in 12.1.1.3 therefore makes an existing destination check effective instead of introducing a new one.

IBM has not published the affected component, the vulnerable endpoint or parameter, the request format, the validation implementation, or the specific bypass technique. API Connect is closed source, so there is no public commit or diff to review. Neither a redirect-based bypass nor DNS rebinding is confirmed as the actual method. We are deliberately not speculating about which check failed.

Severity breakdown

IBM supplies the vector CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:L/A:N. NVD lists IBM as the scoring authority, so the 9.3 is IBM's assessment and not a separate NIST score.

  • Attack vector, Network (AV:N): exploitation occurs over a network path.
  • Attack complexity, Low (AC:L): no special conditions are scored.
  • Privileges required, None (PR:N): no authentication is needed.
  • User interaction, None (UI:N): no other user has to act.
  • Scope, Changed (S:C): impact crosses into a component in a different security scope. Per FIRST, this does not by itself mean complete network compromise.
  • Confidentiality, High (C:H): information exposure is the dominant impact.
  • Integrity, Low (I:L): some integrity impact is scored. IBM does not explain what it consists of.
  • Availability, None (A:N).

Attack flow

Based on IBM's published description, the documented sequence is:

  1. A remote, unauthenticated attacker sends input that causes API Connect to issue an outbound request.
  2. The attacker crafts that input so the product's outbound destination restriction is bypassed.
  3. The API Connect server contacts the unintended destination, for example a host that the server can reach but the attacker cannot.
  4. The server returns the destination's response to the attacker, exposing information through the server instead of through direct access.

The reviewed material provides no reproducible payload, exploit code, or remote code execution chain, and does not specify which secrets could be obtained. The record does not support claims of confirmed credential theft, service disruption, or arbitrary code execution.

Sibling bypasses in the same release

The same release fixes several related SSRF filter bypasses described in similar terms:

  • CVE-2026-17228: destination restrictions on outbound requests can be bypassed
  • CVE-2026-17355: destination checks on outbound requests can be bypassed
  • CVE-2026-17406: address checks can be bypassed
  • CVE-2026-75590: a protection blocking internal addresses can be bypassed

Taken together, these suggest that 12.1.1.3 hardens outbound request validation across several code paths, not just a single endpoint.

Exploitation status

The reviewed sources do not establish active exploitation in the wild. The StackFlag tracker for this CVE lists it as not known to be exploited, and no threat actor, campaign, public proof of concept, or indicators of compromise are identified. The available catalog evidence does not establish CISA KEV membership either way.

Patch information

IBM fixed CVE-2026-17131 in a full product release rather than a standalone hotfix. The security bulletin "Multiple Vulnerabilities in IBM API Connect" (document 7295635, published 5 November 2026) lists the affected product as API Connect V12 OnPrem 12.1.0.0 through 12.1.1.2, and IBM "strongly recommends addressing the vulnerability now by upgrading to v12.1.1.3". NVD gives the same affected range.

  • Vulnerable: 12.1.0.0 through 12.1.1.2
  • Fixed: 12.1.1.3

What the fix addresses

The bug is a bypass of an existing egress filter. The general class of fix for this kind of problem involves:

  • validating the address the request is actually sent to, after DNS resolution and canonicalization, and not the raw user-supplied string
  • rechecking the destination on every redirect
  • treating alternate IP encodings and IPv6-mapped addresses as equivalent to the blocked ranges they represent

These points describe the class of fix, not IBM's specific code change, which has not been published.

Getting the fixed build

The bulletin links to IBM's release page for 12.1.1.3 (support node 7290205), which says the release "delivers important security updates". The files are on IBM Fix Central, require product entitlement, and are dated 3 October 2026. They cover every deployment model:

  • Kubernetes/OpenShift (containers): the API Connect image tool archive and the operator release files for 12.1.1.3.
  • VMware/OVA: subsystem upgrade bundles such as upgrade_management_12.1.1.3.tgz, upgrade_portal_12.1.1.3.tgz, upgrade_devportal_12.1.1.3.tgz and upgrade_analytics_12.1.1.3.tgz, plus full .ova images. Optional appliance control plane files bring Kubernetes to a supported level.
  • Tooling: apicup (Install Assist), apicupgrade and the toolkit for Linux, macOS and Windows. A signature bundle, 12.1.1.3_signatures.zip, is provided to verify downloads.

The release page does not say which subsystem contains the CVE-2026-17131 fix, so upgrade every subsystem (Management, Portal, Analytics and gateways) to 12.1.1.3. A full upgrade is also the reliable way to pick up the sibling SSRF fixes listed above.

Timing and workarounds

The bulletin lists workarounds as "None", so upgrading is the only vendor-supported fix. The fixed build was on Fix Central about a month before the bulletin (5 November 2026), the CVE record (7 November 2026, 04:31:52.553Z) and the NVD entry (7 November 2026, 05:18:26.230). Teams that moved to 12.1.1.3 in October are already covered.

General OWASP SSRF guidance is useful as defense in depth: segment the request functionality, positively allowlist scheme, port and destination, disable redirects in the requesting client, avoid returning raw responses, and log accepted and blocked firewall flows. None of this is an IBM-confirmed substitute for the upgrade, and none of it shows that an affected installation is fixed.

Affected systems and versions

  • Product: IBM API Connect V12 OnPrem
  • Affected versions: 12.1.0.0 through 12.1.1.2, inclusive
  • Fixed version: 12.1.1.3
  • Configuration requirements: none documented. The CVSS vector requires no privileges and no user interaction.

The cited advisory does not establish that other API Connect deployment models or release branches are affected by this CVE.

Vendor security history

IBM handles vulnerabilities through its Product Security Incident Response Team (PSIRT), following a Discovery, Triage, Remediation and Disclosure process influenced by the FIRST PSIRT framework. IBM states that it does not disclose or confirm vulnerabilities until analysis is complete and fixes or mitigations are available, which is consistent with the fixed build predating the bulletin here.

A prior critical issue in the same product is CVE-2025-13915, an API Connect authentication bypass. IBM's bulletin, first published 17 December 2025, describes a remote attacker bypassing authentication mechanisms to gain unauthorized access to the application. Singapore's Cyber Security Agency reported a CVSS 3.1 score of 9.8 and noted that IBM released security updates.

That case and CVE-2026-17131 concern different security boundaries: one is authentication, the other is outbound request destination control. They should not be treated as a single exploit chain or as evidence of a shared root cause. Together they show that critical API Connect issues have been disclosed and addressed through updates. They do not support conclusions about comparative defect rates.

References

Detect & fix
what others miss

Works with
  • GitHub
  • GitLab
  • Bitbucket
  • Azure DevOps Services
  • Gerrit
  • CVS
  • Jira
  • Linear
  • Slack
  • Security Compass
Explore integrations
Security magnifying glass visualization