NVIDIA AIStore Flaw Allows Remote Code Execution in Kubernetes
A critical vulnerability in NVIDIA AIStore for Linux enables remote attackers to execute code and escalate privileges through a legacy API.
Key points
- CVE-2026-65162 affects NVIDIA AIStore framework versions 0 through 4.9.
- The flaw allows remote code execution and privilege escalation via Kubernetes settings.
- The National Vulnerability Database rates the issue as critical with a CVSS score of 10.
NVIDIA AIStore for Linux contains a critical security vulnerability that allows remote attackers to execute arbitrary code. The flaw resides in the legacy ETL Kubernetes PodSpec initialization API, exposing the system to significant risk if exploited by an unauthorized party.
At a glance
- CVE
- CVE-2026-65162
- Product
- NVIDIA AIStore framework
- CVSS
- 10 Critical
- Weakness
- CWE-250
- Exploited
- No known exploitation
According to the National Vulnerability Database, the issue is tracked as CVE-2026-65162. Attackers can supply malicious Kubernetes host-namespace and container-security settings through this legacy interface. A successful exploit may lead to code execution, privilege escalation, information disclosure, data tampering, and denial of service.
Technical Context and Severity
The National Vulnerability Database assigns a CVSS score of 10 to CVE-2026-65162, classifying it as critical. This score reflects the severe potential impact on confidentiality, integrity, and availability of the affected systems. The vulnerability is categorized under CWE-250, which relates to execution with unnecessary privileges.
The specific weakness lies within the legacy ETL Kubernetes PodSpec initialization API. This component is part of the NVIDIA AIStore framework for Linux. By manipulating the initialization process, a remote attacker can inject unauthorized security configurations into Kubernetes pods.
Affected Systems and Scope
The vulnerability affects NVIDIA AIStore framework versions ranging from 0 up to and including version 4.9. Any deployment using these versions of the AIStore for Linux software is potentially vulnerable to this remote code execution flaw. Organizations relying on older builds of the framework are at the highest risk.
The exposure is significant because the attack vector is remote and targets a core initialization component. There is no indication in the source material that a patch or fix has been confirmed or released by NVIDIA for this specific vulnerability yet.
What should you do?
Security teams should immediately audit their environments to identify any installations of NVIDIA AIStore framework versions 0 through 4.9. If these versions are in use, organizations should assess the exposure of the legacy ETL Kubernetes PodSpec initialization API to untrusted networks. Until a vendor-provided update is available, consider restricting network access to the affected API endpoints.
What to do and how to stay safe: NVIDIA
- Inventory all NVIDIA AIStore for Linux installations and verify versions against the affected range of 0 to 4.9.
- Implement network segmentation to restrict remote access to the legacy ETL Kubernetes PodSpec initialization API.
- Monitor system logs for unusual privilege escalation attempts or unexpected container-security setting changes.
- Await official vendor guidance and updates regarding CVE-2026-65162 before applying any unofficial mitigations.
Step-by-step guide: Risk Based Vulnerability Management: Prioritize Fixes by Actual Threat
General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
What is the CVSS score for CVE-2026-65162?
The National Vulnerability Database rates the vulnerability as critical with a CVSS score of 10.
Which versions of NVIDIA AIStore are affected by this flaw?
NVIDIA AIStore framework versions 0 through 4.9 are affected by this remote code execution vulnerability.
Is there a patch available for CVE-2026-65162?
The source material does not confirm that a patch or fix has been released for this vulnerability yet.
