What is the CVSS score of CVE-2026-24215?

CVE-2026-24215 has a CVSS 3.1 base score of 5.7 (Medium). CVSS vector: CVSS:3.1/AV:N/AC:L/PR:L/UI:R/S:U/C:N/I:N/A:H. EPSS exploitation probability: 0.0%.

Triton Inference Server CVE-2026-24215

EUVD-2026-31053 MEDIUM

Uncontrolled Resource Consumption (CWE-400)

2026-05-20 nvidia

GHSA-f637-j435-xwhv

Nvidia Denial Of Service

5.7

CVSS 3.1 · NVD

Severity by source

NVD PRIMARY

5.7 MEDIUM

AV:N/AC:L/PR:L/UI:R/S:U/C:N/I:N/A:H

Primary rating from NVD · only source for this CVE.

CVSS VectorNVD

CVSS:3.1/AV:N/AC:L/PR:L/UI:R/S:U/C:N/I:N/A:H

Attack Vector

Network

Attack Complexity

Low

Privileges Required

Low

User Interaction

Required

Scope

Unchanged

Confidentiality

None

Integrity

None

Availability

High

Lifecycle Timeline

Analysis Generated

May 20, 2026 - 04:05 vuln.today

DescriptionCVE.org

NVIDIA Triton Inference Server contains a vulnerability in the DALI backend, where an attacker could cause uncontrolled resource consumption. A successful exploit of this vulnerability might lead to denial of service.

AnalysisAI

Uncontrolled resource consumption in NVIDIA Triton Inference Server's DALI backend allows a network-adjacent, low-privileged attacker to exhaust server resources, resulting in denial of service. The vulnerability (CWE-400) is triggered through the DALI data-loading and augmentation backend, requires low privileges and user interaction, and carries a CVSS score of 5.7 (Medium). No public exploit code or CISA KEV listing has been identified at time of analysis, placing this in a monitored-but-not-critical-urgency tier for most deployments.

Technical ContextAI

NVIDIA Triton Inference Server is an open-source inference serving platform supporting multiple ML frameworks. The DALI (Data Loading Library) backend is an NVIDIA-developed accelerated data-loading and preprocessing pipeline used within Triton to handle input data for inference requests. CWE-400 (Uncontrolled Resource Consumption) indicates the root cause is a failure to impose appropriate limits on resource usage - such as CPU cycles, memory, threads, or file handles - when processing attacker-controlled input through the DALI pipeline. The affected scope per CPE (cpe:2.3:a:nvidia:triton_inference_server:*:*:*:*:*:*:*:*) covers all versions of the server application without a defined upper or lower version bound in the available data.

RemediationAI

Patch available per vendor advisory - consult the NVIDIA support article at https://nvidia.custhelp.com/app/answers/detail/a_id/5828 for the specific patched release version, as no exact fix version was included in the available input data. In the interim, if the DALI backend is not required for current inference workloads, disabling or restricting access to DALI-backed model endpoints reduces the attack surface without impacting non-DALI model serving. Access controls should be reviewed to ensure that only trusted, explicitly authorized users hold credentials to submit inference requests against the affected server. Rate-limiting and resource quotas at the infrastructure layer (e.g., Kubernetes resource limits, reverse proxy request throttling) can serve as compensating controls to bound the impact of a resource exhaustion attempt, though these do not eliminate the underlying vulnerability. Network segmentation to prevent untrusted or low-trust users from reaching Triton's API port is recommended as an additional defense layer.