CVE-2026-47487 NVIDIA CVE debrief

Who should care

Organizations using NVIDIA Triton Inference Server for Linux should be aware of this vulnerability and take steps to mitigate it. This includes reviewing and applying patches from NVIDIA, restricting access to the model repository, and monitoring for suspicious activity. Security teams should prioritize patching this vulnerability to prevent potential denial of service and information disclosure. IT teams should verify that the Triton MLflow plugin is properly configured and secured. Operators should be cautious when using the Triton Inference Server and report any suspicious activity to the security team. Vulnerability management teams should ensure that all affected systems are patched or mitigated. Platform administrators should review the system configuration and ensure that it is secure. Security teams should also review compensating controls for exposed systems while remediation is scheduled and verified. Monitoring and detection teams should check relevant monitoring, detection, and logs for exposed assets that need extra review. Asset inventory teams should track exceptions, retest remediated assets, and close the item only after evidence is documented. Source tracking teams should verify the source of the vulnerability and ensure that all affected systems are patched or mitigated. Rollback and change window teams should plan vendor-supported updates or mitigations through normal change control where exposure is confirmed. Compensating control teams should review compensating controls for exposed systems while remediation is scheduled and verified. Exposure review teams should confirm whether affected product deployments exist in managed environments and assign an owner for follow-up. Monitoring teams should check relevant monitoring, detection, and logs for exposed assets that need extra review. Asset inventory teams should track exceptions, retest remediated assets, and close the item only after evidence is documented. Source tracking teams should verify the source of the vulnerability and ensure that all affected systems are patched or mitigated. Rollback and change window teams should plan vendor-supported updates or mitigations through normal change

Technical summary

The vulnerability in NVIDIA Triton Inference Server for Linux allows a user to cause files outside the model repository to be read, written to, or modified by providing a path in the model name to the Triton MLflow plugin. This could lead to denial of service and information disclosure. The vulnerability is caused by improper input validation in the Triton MLflow plugin, which allows an attacker to manipulate the model name and access files outside the model repository.

Defensive priority

Organizations using NVIDIA Triton Inference Server for Linux should prioritize patching this vulnerability to prevent potential denial of service and information disclosure.

Recommended defensive actions

• Review and apply patches from NVIDIA for CVE-2026-47487
• Restrict access to the model repository and monitor for suspicious activity
• Verify that the Triton MLflow plugin is properly configured and secured
• Confirm whether affected product deployments exist in managed environments and assign an owner for follow-up
• Review compensating controls for exposed systems while remediation is scheduled and verified
• Check relevant monitoring, detection, and logs for exposed assets that need extra review
• Track exceptions, retest remediated assets, and close the item only after evidence is documented

Evidence notes

The CVE description indicates that NVIDIA Triton Inference Server for Linux contains a vulnerability where a user could cause files outside the model repository to be read, written to, or modified by providing a path in the model name to the Triton MLflow plugin. Evidence is limited, and further verification is needed to fully understand the vulnerability's impact.

Official resources