SB2026051885 - Multiple vulnerabilities in NVIDIA Triton Inference Server (March 2026)

Description

This security bulletin contains information about 3 vulnerabilities.

1) Race condition (CVE-ID: CVE-2025-33238)

CWE-ID: CWE-362 - Concurrent Execution using Shared Resource with Improper Synchronization ('Race Condition')

CVSSv4: CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N/E:U/U:Green

The vulnerability allows a remote attacker to cause a denial of service.

The vulnerability exists due to concurrent execution using shared resource with improper synchronization in the Sagemaker HTTP server when handling requests. A remote attacker can trigger an exception to cause a denial of service.

2) Race condition (CVE-ID: CVE-2025-33254)

CWE-ID: CWE-362 - Concurrent Execution using Shared Resource with Improper Synchronization ('Race Condition')

CVSSv4: CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N/E:U/U:Green

The vulnerability allows a remote attacker to cause a denial of service.

The vulnerability exists due to concurrent execution using shared resource with improper synchronization in Triton Inference Server when handling requests. A remote attacker can cause internal state corruption to cause a denial of service.

3) Uncontrolled Memory Allocation (CVE-ID: CVE-2026-24158)

CWE-ID: CWE-789 - Uncontrolled Memory Allocation

CVSSv4: CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N/E:U/U:Green

The vulnerability allows a remote attacker to cause a denial of service.

The vulnerability exists due to uncontrolled memory allocation in the HTTP endpoint when processing a large compressed payload. A remote attacker can provide a large compressed payload to cause a denial of service.

Remediation

Install update from vendor's website.

References

• https://nvidia.custhelp.com/app/answers/detail/a_id/5790
• https://www.nvidia.com/security/