NVIDIA AIStore for Linux contains a flaw in the legacy ETL Kubernetes PodSpec initialization API, allowing a remote attacker to supply arbitrary host-namespace and container-security settings. The impacted API processes these values without adequate validation, which can lead to code execution, privilege escalation, information disclosure, data tampering, and denial of service. The weakness is classified as CWE‑250, indicating failure to enforce authorization checks.

The vulnerability affects the NVIDIA AIStore framework running on Linux distributions. No specific product versions were listed, so the issue likely applies to all current installations that still expose the legacy ETL Kubernetes API.

The CVSS base score of 10 indicates the highest severity. EPSS data is not available, so the exact likelihood of exploitation cannot be quantified, but the flaw being in a Kubernetes API that can be accessed remotely means attackers possessing API access could exploit it. Since the vulnerability is not yet listed in the CISA KEV catalog, it has not been confirmed as a known, actively exploited vulnerability, but the high severity and remote nature demand immediate attention. Exploitation would typically require the attacker to send crafted PodSpec objects to the AIStore service, potentially granting elevated privileges or execution of arbitrary code on the host.