What is the CVSS score of CVE-2026-47629?

CVE-2026-47629 has a CVSS 3.1 base score of 7.5 (High). CVSS vector: CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H. EPSS exploitation probability: 0.3%.

Which versions of Triton Inference Server are affected by CVE-2026-47629?

NVIDIA Triton Inference Server on Linux contains a remote denial-of-service vulnerability (CVE-2026-47629, CVSS 7.5) accessible to unauthenticated attackers on networked deployments.

How to fix or mitigate CVE-2026-47629?

Within 24 hours: Inventory all NVIDIA Triton Inference Server instances, identify which are reachable from untrusted networks, and document their current versions. Within 7 days: implement network segmentation to restrict access to authorized users only, deploy rate limiting on inference endpoints, and add input validation at the application level to reject malformed requests. Within 30 days: monitor NVIDIA security advisories for a vendor patch release, evaluate containerized or managed inference service options that can simplify future patching, and reassess risk once patch availability is confirmed.

Triton Inference Server CVE-2026-47629

EUVD-2026-61110 HIGH

Improper Input Validation (CWE-20)

2026-08-18 nvidia

GHSA-4hj7-x739-r9gh

Denial Of Service Triton Inference Server Nvidia

7.5

CVSS 3.1 · Vendor: nvidia

Severity by source

Vendor (nvidia) PRIMARY

7.5 HIGH

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

vuln.today AI

7.5 HIGH

Network-accessible inference endpoints require no authentication; impact is strictly availability loss with no confidentiality or integrity compromise possible from a DoS.

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

4.0 AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N

Primary rating from Vendor (nvidia).

CVSS VectorVendor: nvidia

Attack Vector

Network

Attack Complexity

Low

Privileges Required

None

User Interaction

None

Scope

Unchanged

Confidentiality

None

Integrity

None

Availability

High

Lifecycle Timeline

Analysis Generated

Aug 18, 2026 - 18:50 vuln.today

CVE Published

Aug 18, 2026 - 18:24 cve.org

HIGH 7.5

DescriptionCVE.org

NVIDIA Triton Inference Server for Linux contains a vulnerability where an attacker could cause improper input validation. A successful exploit might lead to denial of service.

Articles & Coverage 1

News 6 New Nvidia Vulnerabilities - 1 Critical, 5 High Severity Aug 19, 2026

AnalysisAI

Denial of service in NVIDIA Triton Inference Server for Linux exposes AI inference infrastructure to remote disruption through improper input validation. Unauthenticated network attackers can crash or render unresponsive the inference server by submitting malformed inputs, with CVSS 7.5 (AV:N/AC:L/PR:N/UI:N) confirming no authentication or user interaction is required. No public exploit code has been identified at time of analysis, and CISA KEV listing is absent, but the low attack complexity and zero-authentication requirement make this a meaningful availability risk for any Triton deployment reachable from untrusted networks.

Technical ContextAI

NVIDIA Triton Inference Server is an open-source model serving platform designed to deploy trained AI models at scale across CPU and GPU infrastructure on Linux. It exposes HTTP and gRPC endpoints for inference requests, making it a network-accessible service. The root cause is CWE-20 (Improper Input Validation): the server fails to adequately sanitize or bound-check incoming request data before processing, allowing a specially crafted payload to trigger abnormal server behavior. The CPE string cpe:2.3:a:nvidia:triton_inference_server:*:*:*:*:*:*:*:* with a wildcard version field indicates all currently tracked versions of the server on Linux are potentially within scope, though no upper or lower version bound is confirmed in the available data.

RemediationAI

Consult the NVIDIA Product Security advisory at https://github.com/NVIDIA/product-security/tree/main/2026/5865 for the authoritative patch guidance; no exact fixed version number was present in the provided intelligence data, so a specific upgrade target cannot be confirmed independently. Until a patch is applied, compensating controls should include restricting network access to Triton's HTTP (default port 8000) and gRPC (default port 8001) endpoints to trusted internal IP ranges using firewall rules or network policies, with the trade-off that this may block legitimate inference clients. Placing a reverse proxy or API gateway with input size limits and request rate limiting in front of Triton can reduce the attack surface without disabling the service, though it does not eliminate the underlying validation flaw. Monitoring inference endpoint error rates and process restart frequency can provide early warning of exploitation attempts.