IBM i Uncontrolled Search Path Privilege Escalation (CVE-2026-16860)

Technical Analysis

The vulnerability stems from an uncontrolled search path element (CWE-427) in IBM i, where the operating system or an associated component resolves executables or libraries using a search order that can be manipulated by an authenticated attacker. By planting a malicious binary or library earlier in the search path, an attacker can achieve arbitrary code execution in the context of the vulnerable process, likely leading to privilege escalation given the 9.9 CVSS score and lack of required privileges beyond authentication. This attack vector is common in legacy midrange server environments and requires only network access and valid credentials, no user interaction. Organizations running AI agent orchestration, RAG pipelines, or LLM tool-calling backends on IBM i hosts (e.g., for ERP integration, database connectivity via Db2 for i, or legacy business logic invoked by agents) face risk of credential theft, agent tool-chain compromise, or lateral movement if this system is part of an agentic workflow's backend infrastructure.

Affected Systems

IBM i 7.3, 7.4, 7.5, and 7.6 (all supported releases at time of disclosure); specific exposure depends on which IBM i components or licensed programs load executables/libraries via the vulnerable search path mechanism

Indicators of Compromise

• No specific IOCs published at this time; monitor for unauthorized executable/library placement in IBM i library lists (*LIBL) and system directories, and unexpected process execution under elevated authorities (e.g., *ALLOBJ, *SECADM users)

Remediation Steps

1. 1

   Apply IBM PTF/Security Fix

   Obtain and install the official IBM Program Temporary Fix (PTF) addressing CVE-2026-16860 for your specific IBM i release (7.3, 7.4, 7.5, or 7.6) as soon as it is available from IBM Fix Central.

2. 2

   Restrict Authenticated Access

   Limit the number of users with authenticated access to IBM i systems and enforce least-privilege principles, particularly for accounts capable of writing to shared library paths or directories.

3. 3

   Audit Library Lists and Search Paths

   Review and lock down system and user library lists (*LIBL) to prevent unauthorized insertion of malicious objects ahead of trusted system libraries.

4. 4

   Monitor for Exploitation Attempts

   Enable auditing (QAUDJRN) for object creation and program execution events, and alert on anomalous library or executable creation by non-administrative accounts.

5. 5

   Segment IBM i from Agent Infrastructure

   If IBM i systems interface with AI agent pipelines or automation tools, ensure network segmentation and credential isolation so that compromise of the IBM i host cannot directly expose agent API keys or orchestration credentials.