CVE-2026-12728: IBM MQ Deserialization RCE Vulnerability

CVE-2026-12728 Overview

CVE-2026-12728 is a deserialization of untrusted data vulnerability [CWE-502] affecting multiple versions of IBM MQ. An authenticated attacker can send crafted serialized objects to the message queuing service to execute arbitrary code on the target system. The flaw impacts IBM MQ Long Term Support (LTS) and Continuous Delivery (CD) branches across the 9.1, 9.2, 9.3, 9.4, and 10.0 release trains.

Critical Impact

Authenticated attackers can achieve arbitrary code execution on IBM MQ hosts, compromising message broker confidentiality, integrity, and availability across enterprise messaging infrastructure.

Affected Products

• IBM MQ 9.1.0.0 through 9.1.0.37 LTS and 9.2.0.0 through 9.2.0.43 LTS
• IBM MQ 9.3.0.0 through 9.3.0.41 LTS, 9.3.0.0 through 9.3.5.1 CD, 9.4.0.0 through 9.4.0.25 LTS, and 9.4.0.0 through 9.4.5.1 CD
• IBM MQ 10.0.0.0

Discovery Timeline

• 2026-09-15 - CVE-2026-12728 published to NVD
• 2026-09-17 - Last updated in NVD database

Technical Details for CVE-2026-12728

Vulnerability Analysis

The vulnerability resides in IBM MQ's handling of serialized Java objects transmitted through the messaging service. When the broker deserializes untrusted input without adequate type filtering, an attacker can supply a crafted object graph that triggers unintended code paths during reconstruction. Successful exploitation results in arbitrary code execution in the context of the MQ process.

Exploitation requires prior authentication to the MQ instance, which limits opportunistic attacks but does not prevent abuse by compromised service accounts or insider threats. The vulnerability affects both LTS and CD release trains, indicating the flawed deserialization logic exists in a shared code path across the product family.

Root Cause

The root cause is insecure deserialization [CWE-502] of untrusted data. IBM MQ processes serialized objects from authenticated clients without enforcing a strict allow-list of permitted classes. Gadget chains present in the classpath can be leveraged during object reconstruction to invoke sensitive methods, culminating in arbitrary command execution.

Attack Vector

The attack is delivered over the network to the MQ listener port. An authenticated attacker crafts a malicious serialized payload and submits it through a supported client interface. When the queue manager deserializes the payload, embedded gadget chains execute attacker-controlled logic. No user interaction is required beyond initial authentication, and the impact spans confidentiality, integrity, and availability.

See the IBM Security Patch Information advisory for technical details specific to the affected components.

Detection Methods for CVE-2026-12728

Indicators of Compromise

• Unexpected child processes spawned by the IBM MQ queue manager (amqzmuc0, amqzxma0, runmqlsr) invoking shells, scripting interpreters, or curl/wget.
• Anomalous outbound network connections initiated by the MQ service account following receipt of large or malformed serialized messages.
• MQ error logs (AMQERR01.LOG) containing Java deserialization stack traces or ClassNotFoundException entries referencing suspicious gadget classes.

Detection Strategies

• Inspect MQ message payloads for Java serialization magic bytes (AC ED 00 05) originating from unexpected clients or channels.
• Correlate authentication events on MQ channels with subsequent process creation events on the host to surface post-auth exploitation.
• Enable MQ channel authentication records and audit events to identify unusual client identities interacting with sensitive queues.

Monitoring Recommendations

• Forward MQ error logs, FFST files, and OS process telemetry to a centralized analytics platform for correlation.
• Baseline expected MQ client applications and alert on new or unrecognized client IDs authenticating to the queue manager.
• Monitor listener ports (default 1414) for connection spikes and payload size anomalies indicative of exploitation attempts.

How to Mitigate CVE-2026-12728

Immediate Actions Required

• Apply the fixed IBM MQ maintenance level referenced in the vendor advisory to all LTS and CD deployments.
• Rotate credentials for MQ service accounts and administrative users to limit reuse of exposed authentication material.
• Restrict network access to MQ listener ports so only trusted application hosts can reach the queue manager.

Patch Information

IBM has released fixes for affected LTS and CD versions. Consult the IBM Security Patch Information bulletin for the specific fix pack, iFix, or CSU required for each supported release stream.

Workarounds

• Enforce channel authentication records (CHLAUTH) to block anonymous or low-privilege identities from reaching sensitive queues.
• Apply strict authorization on SYSTEM.* queues and administrative channels using setmqaut to reduce the pool of accounts capable of triggering the flaw.
• Segment MQ hosts on isolated network zones with firewall policies that limit inbound connections to vetted application servers.

# Configuration example: restrict MQ channel access with CHLAUTH
runmqsc QMGR1 <<EOF
SET CHLAUTH('*') TYPE(ADDRESSMAP) ADDRESS('*') USERSRC(NOACCESS) ACTION(REPLACE)
SET CHLAUTH('APP.SVRCONN') TYPE(ADDRESSMAP) ADDRESS('10.10.20.0/24') USERSRC(CHANNEL) ACTION(REPLACE)
SET CHLAUTH('*') TYPE(BLOCKUSER) USERLIST('*MQADMIN') ACTION(REPLACE)
REFRESH SECURITY TYPE(CONNAUTH)
EOF