IBM i CVE-2026-16826

Exploitation requires a valid local account on the affected IBM i system with sufficient authority to invoke the vulnerable component (PR:L per CVSS vector). The attack vector is strictly local (AV:L), meaning direct network exploitation is not possible; a remote attacker would first need to establish a foothold on the system. No user interaction is required (UI:N) beyond the attacker's own actions. The specific IBM i subsystem, utility, or command interface containing the injection point is not identified in publicly available data, which limits opportunistic exploitation. Exploitation is bounded to IBM i releases 7.3, 7.4, 7.5, and 7.6 and does not affect other IBM platforms. No special non-default configuration is identified as a prerequisite.

The CVSS 7.8 score is partially offset by the strictly local attack vector (AV:L), which prevents direct remote exploitation without a prior foothold. The EPSS score of 0.12% at the 2nd percentile is notably low, indicating no observed exploitation activity and minimal current threat-actor interest. There is no CISA KEV entry and no known public proof-of-concept. IBM i tends to be deployed in high-security environments such as banking, insurance, and government, where local user access is tightly controlled - further reducing practical risk in well-managed deployments. However, environments hosting multi-tenant IBM i partitions or granting CL or PASE shell access to semi-trusted personnel face the most meaningful exposure, and those operators should treat this as a priority patch cycle item. The combination of low EPSS, local-only vector, and confirmed patch availability places this in the 'patch on next maintenance window' tier rather than emergency remediation.

A local attacker with a standard IBM i user account supplies crafted input containing OS metacharacters (e.g., a semicolon followed by a privileged command) to the vulnerable IBM i function or utility. The application passes this input unsanitized to an OS command interpreter, executing the injected command under the privileges of the affected process - potentially achieving full system compromise. No public proof-of-concept has been identified, but the low attack complexity (AC:L) means exploitation would be mechanically straightforward once the specific vulnerable call site is located.