CVE-2026-15194: Open5GS 2.7.7 Use-After-Free in AMF Context Handling

Technical summary

CVE-2026-15194 is a use-after-free memory safety defect (CWE-416) combined with out-of-bounds access potential (CWE-119) located in src/amf/context.c within Open5GS 2.7.7's AMF subsystem. The flaw occurs during context finalization when object references are freed but subsequently accessed. The vulnerability requires local system access and authenticated user privileges to exploit. Public proof-of-concept code has been disclosed, removing the barrier to practical weaponization.

Business impact

For operators running Open5GS in lab, test, or pre-production environments, this vulnerability poses a localized information disclosure risk. The requirement for local, authenticated access limits blast radius to insider threats or compromised system accounts. However, in containerized or multi-tenant deployments where AMF instances share kernel namespaces or where unprivileged container users have elevated permissions, the risk surface expands. The availability of public exploits means security teams cannot rely on obscurity as a defense layer.

Affected systems

Open5GS version 2.7.7 is affected. The vulnerability resides specifically in the AMF component's context handling during lifecycle operations. Operators running earlier or later versions should verify their patch status against the vendor advisory, as the scope and fix date of this issue require confirmation from Open5GS maintainers.

Exploitability

Exploitability is moderate in practical terms despite the low CVSS score. The low score reflects the local, authenticated prerequisite—the attacker must already have a shell session or service account on the Open5GS system. Public exploit code now circulates, eliminating the need for custom development. The attack requires no user interaction and is deterministic. In environments where developers, testers, or service accounts have interactive access, exploitation becomes straightforward.

Remediation

Upgrade Open5GS to a patched version released after this vulnerability's disclosure. Consult the Open5GS project repository and security advisories for the specific version number containing the fix. Until patching is feasible, restrict local shell and service account access to the AMF host to authenticated administrators only, disable unnecessary local login methods, and implement file integrity monitoring on the AMF binary and context.c-derived components.

Patch guidance

Contact the Open5GS project maintainers or check their official GitHub releases and security bulletins for the recommended patched version. Apply patches in a staging environment first to validate AMF stability and 5G core functionality. Given the local-only attack vector, patching can typically proceed on a standard maintenance window rather than emergency timelines, but should not be indefinitely deferred given public exploit availability.

Detection guidance

Monitor for segmentation faults or abnormal termination of the amf process with SIGSEGV signals, particularly correlated with user session creation or context cleanup events. Deploy AddressSanitizer (ASan) or Valgrind on development/test AMF instances to catch use-after-free memory accesses during functional testing. Log and alert on failed or suspicious authentication attempts to local AMF service accounts. Memory dump analysis of crashed AMF processes may reveal heap corruption patterns consistent with this flaw.

Why prioritize this

Although CVSS is low (3.3), this vulnerability merits medium priority because: (1) public exploit code exists, eliminating discovery barriers; (2) the local+authenticated requirement is not insurmountable in containerized or shared hosting scenarios; (3) information disclosure via heap read could expose 5G subscription credentials or session keys; and (4) use-after-free bugs can occasionally be chained with other flaws to achieve higher impact. Defer below critical/high issues but schedule patching within 30–60 days.

Risk score, explained

The CVSS 3.1 score of 3.3 (LOW) reflects: Attack Vector=Local (AV:L), Attack Complexity=Low (AC:L), Privileges Required=Low (PR:L), User Interaction=None (UI:N), Scope=Unchanged (S:U), Confidentiality=Low (C:L), Integrity=None (I:N), Availability=None (A:N). The score correctly penalizes the local access requirement and lack of integrity/availability impact. However, it does not account for the public exploit availability, which increases practical exploitability beyond the base score. Organizations should treat this as a local privilege escalation + information disclosure compound risk rather than a purely low-severity issue.

Frequently asked questions