IBM MQ CVE-2026-12355

Exploitation requires an attacker to supply a crafted value that IBM MQ passes to a JNDI lookup without sufficient input validation, AND an attacker-controlled directory service (LDAP/RMI or similar) that is network-reachable from the MQ server so the malicious object reference can be resolved and retrieved. The CVSS AC:H rating captures these compound prerequisites. Limiting factors: successful RCE depends on object retrieval/instantiation succeeding in the MQ JVM (which may be constrained by JDK version and deserialization/trust-URL protections such as com.sun.jndi.*.trustURLCodebase settings), and the description names neither the specific product feature nor endpoint, so operators cannot yet scope which interface (client, JMS admin, or configuration path) is the entry point. Applies to the listed 9.1-9.4 LTS/CD streams and MQ 10.0.0.0; unaffected/patched releases are available from IBM. PR:N indicates no authentication is required per the provided vector.

This is a genuinely serious flaw but not an immediate emergency based on current signals. The CVSS 3.1 vector (AV:N/AC:H/PR:N/UI:N/C:H/I:H/A:H, base 8.1) indicates a network-reachable, unauthenticated (PR:N) JNDI injection with full technical impact, which the SSVC framework corroborates (Technical Impact: total). However, three signals temper the urgency: AC:H reflects real exploitation hurdles (the attacker must both control a value that reaches a JNDI lookup and stand up/reach a malicious directory server), SSVC records Exploitation: none and Automatable: no, and there is no CISA KEV listing, no public POC, and no EPSS score in the input. The overall picture is a high-severity, high-impact vulnerability in widely-deployed enterprise middleware that warrants prompt patching, but there is no evidence of active or automatable exploitation at analysis time - it is a real priority driven by impact and product footprint rather than by observed weaponization. Note the description says RCE is only 'potentially' achievable (disclosure is the confirmed floor), so C:H/I:H/A:H represents worst-case JNDI-to-RCE rather than a proven end-to-end chain.

An attacker submits crafted input to an IBM MQ interface that MQ passes to a JNDI lookup without sufficient validation, causing the server to resolve an attacker-controlled LDAP/RMI reference (Log4Shell-style). The referenced directory server returns a malicious object that, when retrieved and instantiated, leaks sensitive data or executes attacker-supplied code in the MQ process context. No public exploit code is identified at time of analysis, and the AC:H rating reflects the need for a reachable attacker-controlled directory server plus a value that lands in a vulnerable lookup.