← Back to Feed CACHED · 2026-07-31 01:33:19 · CACHE_KEY CVE-2026-14529
CVE-2026-14529 · CWE-306 · Disclosed 2026-07-29

IBM WebSphere Application Server 9.0

ASSESSED — NOISGATE V0.5
Vendor
Reassessed
Verdict:
Do you agree?
01 · The Real Story

Like leaving a master key under the mat of a side door that 95% of buildings never installed

CVE-2026-14529 is a missing-authentication flaw (CWE-306) in the SIP (Session Initiation Protocol) container of IBM WebSphere Application Server. When the sipServlet-1.1 feature is enabled, an unauthenticated remote attacker can forge server-side requests through the SIP endpoint (typically port 5060/5061), achieving high-impact reads and writes against internal services. Affected versions span WAS traditional 8.5.x and 9.0.x, plus Liberty 17.0.0.3 through 26.0.0.8. Fixes target WAS 9.0.5.29 and Liberty 26.0.0.9, both slated for Q3 2026.

IBM rated this CRITICAL at 9.4 and the CVSS vector backs it up on paper — network-reachable, no auth, no user interaction, high confidentiality and integrity impact. That score is *technically correct for the subset of installs running SIP*. But the SIP container is a niche telephony feature used in unified communications and VoIP deployments; the overwhelming majority of WebSphere estates never enable sipServlet-1.1. This prerequisite dramatically shrinks the real-world blast radius. Combined with an EPSS of 0.33%, no KEV listing, no public PoC, and no GreyNoise signals, the vendor score overstates risk for most enterprises. noisgate downgrades to HIGH 7.5 — still urgent for SIP-enabled deployments, but not a fleet-wide fire drill.

"Unauthenticated SSRF is real, but only if you turned on the SIP feature almost nobody uses."
02 · The Attack Path

4 steps from start to impact.

STEP 01

Identify SIP-enabled WebSphere instance

The attacker scans for SIP listener ports (UDP/TCP 5060, TLS 5061) or probes HTTP management endpoints for sipServlet-1.1 feature indicators. WebSphere SIP containers advertise via SIP OPTIONS responses, making fingerprinting straightforward. Shodan or FOFA queries for SIP banners on WebSphere-associated IPs would narrow targets.
Conditions required:
  • Target WebSphere instance has sipServlet-1.1 feature enabled
  • SIP listener port is network-reachable from the attacker's position
Where this breaks in practice:
  • sipServlet-1.1 is a niche feature for VoIP/UC — estimated <5% of WebSphere deployments enable it
  • SIP ports (5060/5061) are rarely exposed to the internet; typically firewalled to voice VLANs
Detection/coverage: Nessus plugin family covers WebSphere version detection; custom SIP port scans via nmap --script sip-methods can identify exposed listeners.
STEP 02

Send unauthenticated SSRF payload

Because the SIP container lacks authentication on a critical function (CWE-306), the attacker crafts a SIP message with a manipulated URI to force the WebSphere server to issue an outbound request to an attacker-controlled or internal destination. No credentials or session tokens are required. The CVSS vector confirms PR:N/UI:N.
Conditions required:
  • Network path to SIP listener is open
  • No WAF or SIP-aware firewall inspecting inbound SIP traffic
Where this breaks in practice:
  • SIP-aware firewalls (session border controllers) in telecom deployments will validate SIP message structure
  • Enterprise network segmentation typically isolates voice infrastructure
Detection/coverage: SIP proxy logs or IDS rules for anomalous SIP INVITE/OPTIONS with non-standard Request-URIs.
STEP 03

Pivot to internal services via SSRF

The forged server-side request reaches internal endpoints — metadata services (cloud IMDSv1 at 169.254.169.254), admin consoles, database ports, or other application tiers. The attacker reads responses (C:H) and can write/modify state on services that trust the WebSphere server's source IP (I:H). The Scope is Unchanged, meaning impact is confined to the WebSphere server's own authorization context.
Conditions required:
  • Internal services reachable from the WebSphere host
  • Target internal services trust requests from the WebSphere server IP or lack their own authentication
Where this breaks in practice:
  • Cloud deployments with IMDSv2 enforce PUT-based token exchange, blocking simple GET SSRF
  • Zero-trust or microsegmented networks limit lateral reach from the app tier
  • Internal services with their own authentication layers reduce write impact
Detection/coverage: Cloud provider IMDS access logs, WAF/proxy logs showing unusual outbound requests from the WebSphere host, network flow anomaly detection.
STEP 04

Data exfiltration or integrity compromise

With SSRF access to internal services, the attacker exfiltrates sensitive data (credentials, API keys, configuration) or modifies application state. In cloud environments, leaked IAM credentials from metadata services can escalate to broader account compromise. Impact is bounded by what internal services the WebSphere host can reach and what those services authorize.
Conditions required:
  • Valuable data or credentials accessible via SSRF-reachable endpoints
  • No egress filtering preventing data exfiltration
Where this breaks in practice:
  • Modern cloud deployments rotate short-lived credentials, limiting exploitation windows
  • DLP and egress monitoring can detect anomalous outbound data flows
Detection/coverage: SIEM correlation of unusual outbound connections from WebSphere hosts, cloud audit trails for credential usage from unexpected sources.
03 · Intelligence Metadata

The supporting signals.

In-the-wild exploitationNot observed. No CISA KEV listing, no GreyNoise tags, no vendor acknowledgment of active exploitation as of 2026-07-31.
Proof-of-conceptNone public. No GitHub repos, no researcher writeups, no exploit framework modules identified. The predecessor CVE-2025-27907 (similar SIP SSRF, but requiring high privileges) also had no public PoC.
EPSS0.00333 (0.33%) — bottom quartile. The exploitation probability model sees this as unlikely to be weaponized in the near term.
KEV statusNot listed on CISA KEV as of 2026-07-31.
CVSS vectorCVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:L — The vector is maximally accessible (network, no auth, no interaction) but Scope:Unchanged limits blast radius to the vulnerable component's context.
Affected versionsWAS traditional: 8.5.x (all fix packs), 9.0.0.0–9.0.5.28. Liberty: 17.0.0.3–26.0.0.8. Requires sipServlet-1.1 feature enabled.
Fixed versionsLiberty: 26.0.0.9 (targeted Q3 2026). WAS traditional 9.0: Fix Pack 9.0.5.29 (targeted Q3 2026). WAS 8.5: Check IBM support for interim fix (APAR TBD). Interim fixes likely available before fix packs.
Predecessor CVECVE-2025-27907 — same SIP container SSRF pattern, but rated Medium 4.1 because it required high privileges (PR:H). This new CVE removes the auth requirement, escalating severity significantly.
Scanning/exposureNo Shodan or Censys data specific to WebSphere SIP endpoints. SIP ports (5060/5061) are rarely internet-facing in enterprise deployments. WebSphere admin consoles on 9043/9060 are more commonly scanned but are a different attack surface.
Disclosure timelineReserved 2026-07-02, published 2026-07-29, updated 2026-07-30. Two-day-old disclosure with no exploitation signals.
04 · The Call

noisgate verdict.

Final Verdict
DOWNGRADED to HIGH (7.5/10)

The single most decisive factor driving the downgrade is the narrow affected population: exploitation requires the sipServlet-1.1 feature to be enabled, a niche VoIP/telephony capability that fewer than 5% of WebSphere deployments use, compounded by SIP ports being almost never internet-exposed. For the small subset with SIP enabled, the unauthenticated network-reachable vector with high C/I impact prevents any further downgrade below HIGH.

HIGH Vulnerability characteristics and CVSS interpretation
MEDIUM Affected population estimate (sipServlet-1.1 usage is niche but no hard telemetry)
HIGH No active exploitation or public PoC

Why this verdict

  • sipServlet-1.1 prerequisite narrows population: The SIP container feature is a specialized telephony/UC capability. The vast majority of WebSphere deployments are traditional Java EE application servers that never enable SIP. This prerequisite alone eliminates ~95% of the installed base from being vulnerable, warranting a meaningful severity reduction.
  • No exploitation signals: EPSS at 0.33%, no KEV listing, no GreyNoise tags, no public PoC, and the CVE is only two days old. The predecessor CVE-2025-27907 (same component, auth-required variant) never saw exploitation either. The practical threat is currently theoretical.
  • SIP ports rarely internet-facing: Enterprise SIP infrastructure typically sits behind session border controllers (SBCs) on internal voice VLANs. Even when sipServlet-1.1 is enabled, the SIP listener is almost never directly reachable from the internet, adding a network-access friction layer the CVSS score doesn't account for.
  • Role multiplier: WebSphere is an application server (typical role: line-of-business tier). The SIP-enabled subset serves telephony workloads — not identity, not backup, not hypervisor. Chain success in a SIP-enabled deployment yields SSRF from the app tier (host-level blast radius, possibly tenant-level if cloud IMDS is reachable). This does not reach domain/fleet/supply-chain scale. The role does not trigger the HIGH→CRITICAL floor override.
  • Scope:Unchanged limits blast radius: The CVSS S:U designation means impact is confined to the WebSphere server's own authorization context. The attacker does not break out of the component boundary, though SSRF can indirectly reach other services the host can access.

Why not higher?

The CVSS vector is maximally accessible (AV:N/AC:L/PR:N/UI:N) and the C:H/I:H impact is real when SIP is enabled. However, CRITICAL requires either broad population exposure or a high-value-role multiplier. The sipServlet-1.1 prerequisite eliminates the broad population condition, and WebSphere SIP endpoints are not canonically fleet-scale infrastructure (not DCs, not hypervisors, not identity providers). No active exploitation further undermines urgency.

Why not lower?

Despite the narrow affected population, this is still an unauthenticated SSRF with missing authentication on a critical function — the most dangerous SSRF class. For organizations that DO run SIP on WebSphere, the chain is trivial (no auth, low complexity) and impact is high. The predecessor CVE-2025-27907 was only Medium because it required admin-level auth; removing that gate is a material escalation. MEDIUM would understate the risk for affected deployments.

05 · Compensating Control

What to do — in priority order.

  1. Disable sipServlet-1.1 if not actively used — Most WebSphere deployments do not need SIP functionality. In Liberty, remove sipServlet-1.1 from server.xml features. In WAS traditional, remove SIP container configuration. This eliminates the vulnerability entirely. Verify within 30 days per the noisgate mitigation SLA for HIGH findings.
  2. Firewall SIP ports to authorized voice VLANs only — Block inbound traffic to ports 5060/5061 (UDP and TCP) from all sources except known SIP peers (IP-PBX, SBC, UC clients). Apply at both host firewall and network perimeter. This prevents unauthenticated remote exploitation even if the feature remains enabled. Deploy within 30 days.
  3. Deploy a session border controller (SBC) in front of SIP endpoints — SBCs validate SIP message structure, enforce topology hiding, and can block malformed or suspicious SIP URIs before they reach WebSphere. If you run production SIP, you should already have an SBC — verify it is inline. Deploy within 30 days.
  4. Enable IMDSv2 on cloud-hosted WebSphere instances — If WebSphere runs on AWS EC2, enforcing IMDSv2 (hop limit=1, PUT token required) blocks the most common SSRF-to-credential-theft chain. This limits the blast radius of any SSRF exploitation. Deploy immediately for cloud workloads.
What doesn't work
  • Web Application Firewall (WAF): WAFs inspect HTTP traffic on ports 80/443. SIP traffic on 5060/5061 uses a different protocol; HTTP-layer WAFs will not see or filter it.
  • WebSphere admin console hardening: Restricting admin console access (ports 9043/9060) does not affect the SIP container listener, which runs on entirely separate ports.
  • Upgrading Java SDK alone: The vulnerability is in the WebSphere SIP container code, not the JVM. Patching the IBM SDK for Java will not remediate this CVE.
06 · Verification

Crowdsourced verification payload.

Run on each WebSphere host (traditional or Liberty) as the WebSphere admin user or root. For Liberty: bash check_cve_2026_14529.sh /opt/ibm/wlp. For traditional WAS: bash check_cve_2026_14529.sh /opt/IBM/WebSphere/AppServer. No network access required.

noisgate-verify.sh
BASHREAD-ONLYSAFE
#!/bin/bash
# check_cve_2026_14529.sh — Detect CVE-2026-14529 exposure
# Usage: bash check_cve_2026_14529.sh <WAS_HOME>
# Exit codes: 0=PATCHED, 1=VULNERABLE, 2=UNKNOWN

set -euo pipefail
WAS_HOME="${1:-}"

if [ -z "$WAS_HOME" ]; then
  echo "Usage: $0 <WAS_HOME>"
  echo "UNKNOWN - no WAS_HOME provided"
  exit 2
fi

SIP_ENABLED=false
IS_LIBERTY=false
RESULT="UNKNOWN"

# Detect Liberty vs traditional
if [ -f "$WAS_HOME/bin/server" ] || [ -d "$WAS_HOME/usr/servers" ]; then
  IS_LIBERTY=true
  echo "[*] Detected Liberty installation at $WAS_HOME"
  # Check all server.xml files for sipServlet-1.1
  while IFS= read -r -d '' sxml; do
    if grep -qi 'sipServlet-1.1' "$sxml" 2>/dev/null; then
      SIP_ENABLED=true
      echo "[!] sipServlet-1.1 ENABLED in: $sxml"
    fi
  done < <(find "$WAS_HOME/usr/servers" -name 'server.xml' -print0 2>/dev/null)
  
  # Check Liberty version
  VERSION_FILE="$WAS_HOME/lib/versions/openliberty.properties"
  [ ! -f "$VERSION_FILE" ] && VERSION_FILE="$WAS_HOME/lib/versions/WebSphereApplicationServer.properties"
  if [ -f "$VERSION_FILE" ]; then
    VERSION=$(grep 'product.version' "$VERSION_FILE" 2>/dev/null | head -1 | cut -d'=' -f2 | tr -d ' ')
    echo "[*] Liberty version: $VERSION"
    # Compare: vulnerable if < 26.0.0.9
    if [ -n "$VERSION" ]; then
      MAJOR=$(echo "$VERSION" | cut -d. -f1)
      MINOR=$(echo "$VERSION" | cut -d. -f2)
      PATCH=$(echo "$VERSION" | cut -d. -f3)
      BUILD=$(echo "$VERSION" | cut -d. -f4)
      if [ "$MAJOR" -lt 26 ] || ([ "$MAJOR" -eq 26 ] && [ "$MINOR" -eq 0 ] && [ "$PATCH" -eq 0 ] && [ "$BUILD" -lt 9 ]); then
        if $SIP_ENABLED; then
          RESULT="VULNERABLE"
        else
          RESULT="PATCHED"
          echo "[*] Version is in affected range but sipServlet-1.1 is NOT enabled — not exploitable"
        fi
      else
        RESULT="PATCHED"
      fi
    fi
  else
    echo "[?] Could not determine Liberty version"
  fi
else
  echo "[*] Detected traditional WAS installation at $WAS_HOME"
  # Check for SIP container in traditional WAS
  if [ -x "$WAS_HOME/bin/versionInfo.sh" ]; then
    VERSION=$($WAS_HOME/bin/versionInfo.sh 2>/dev/null | grep 'Version' | head -1 | awk '{print $NF}')
    echo "[*] WAS version: $VERSION"
  fi
  # Check SIP in config
  CONFIGS=$(find "$WAS_HOME/profiles" -name 'server.xml' -o -name 'sip*.xml' 2>/dev/null || true)
  for cfg in $CONFIGS; do
    if grep -qi 'sipcontainer\|sipservlet\|SIPContainer' "$cfg" 2>/dev/null; then
      SIP_ENABLED=true
      echo "[!] SIP container reference found in: $cfg"
    fi
  done
  if [ -n "${VERSION:-}" ]; then
    FP=$(echo "$VERSION" | cut -d. -f4)
    MAJOR=$(echo "$VERSION" | cut -d. -f1)
    MINOR=$(echo "$VERSION" | cut -d. -f2)
    if ([ "$MAJOR" -eq 9 ] && [ "${FP:-0}" -lt 29 ]) || [ "$MAJOR" -eq 8 ]; then
      if $SIP_ENABLED; then
        RESULT="VULNERABLE"
      else
        RESULT="PATCHED"
        echo "[*] Version is in affected range but SIP container is NOT enabled — not exploitable"
      fi
    else
      RESULT="PATCHED"
    fi
  fi
fi

if ! $SIP_ENABLED && [ "$RESULT" != "PATCHED" ]; then
  echo "[*] sipServlet-1.1 / SIP container not detected in configuration"
  RESULT="PATCHED"
fi

echo ""
echo "=== CVE-2026-14529 STATUS: $RESULT ==="
exit $([ "$RESULT" = "PATCHED" ] && echo 0 || ([ "$RESULT" = "VULNERABLE" ] && echo 1 || echo 2))
07 · Bottom Line

If you remember one thing.

TL;DR
Monday morning: First, run the verification script across your WebSphere fleet to identify which instances have sipServlet-1.1 enabled — this is your actual affected population, likely a small fraction of your estate. For any SIP-enabled instances, disable the feature immediately if it is not required for active telephony workloads. For instances where SIP is production-critical, firewall ports 5060/5061 to authorized SIP peers only and place a session border controller inline. Per the noisgate mitigation SLA for HIGH findings, compensating controls must be in place within 30 days. Apply the vendor patch (Liberty 26.0.0.9 or WAS 9.0.5.29, expected Q3 2026) within the 180-day noisgate remediation SLA. If IBM releases an interim fix or APAR before the fix pack, prefer that. There is no active exploitation or public PoC today, so this is a controlled remediation, not a crisis — but do not let the narrow population lull you into forgetting the SIP boxes entirely.

Sources

  1. CVE-2026-14529 — THREATINT
  2. IBM Security Bulletin — WebSphere Multiple Vulnerabilities (July 2026)
  3. HKCERT — IBM WebSphere Products Multiple Vulnerabilities (2026-07-30)
  4. CVE-2025-27907 — Predecessor SIP SSRF (NVD)
  5. IBM Documentation — sipServlet-1.1 Feature
  6. IBM Documentation — SIP in WebSphere Application Server
  7. CWE-306 — Missing Authentication for Critical Function
Peer Review

What defenders are saying.

Submit a review attribution: handle + country only
0 flags selected · stored anonymously
Validation Results

Crowdsourced verification outputs.

Results submitted by users who ran the verification payload against their environment.