CVE-2026-47628: 
Triton Inference Server vulnerability analysis and mitigation

CVE-2026-47628 is a denial-of-service vulnerability in NVIDIA Triton Inference Server for Linux caused by allocation of resources without limits or throttling (CWE-770). It was published on August 18, 2026, as part of NVIDIA's Security Bulletin 5865, which covers multiple CVEs affecting Triton Inference Server. The affected version is listed as 0.0-26.05 (up to and including the 26.05 release). It carries a CVSS v3.1 base score of 7.5 (High), assigned by NVIDIA Corporation (NVIDIA Advisory).

The vulnerability is classified under CWE-770 (Allocation of Resources Without Limits or Throttling), meaning the server fails to impose adequate caps on resource consumption triggered by incoming requests. An unauthenticated remote attacker can exploit this over the network with low complexity and no user interaction required, by sending crafted requests that cause unbounded resource allocation — consistent with attack patterns such as excessive allocation (CAPEC-130) or application exhaustion flooding (T1499.003). No public proof-of-concept code or detailed technical write-up has been disclosed at this time (NVIDIA Advisory).

Successful exploitation results in a denial-of-service condition on the affected Triton Inference Server instance, with high availability impact and no confidentiality or integrity impact. Because Triton Inference Server is commonly used to serve AI/ML inference workloads in production environments, an outage could disrupt downstream applications and services relying on model inference. The CISA SSVC assessment characterizes the technical impact as partial and the vulnerability as automatable, indicating it could be triggered at scale without manual effort (NVIDIA Advisory).

According to CISA's SSVC assessment, there is currently no known exploitation of this vulnerability in the wild, and it has not been added to the CISA Known Exploited Vulnerabilities (KEV) catalog. The vulnerability is rated as automatable, meaning exploitation can be scripted without human interaction. The EPSS score is 0.0 at time of publication, reflecting low current exploitation probability. No exploit kits, weaponized tools, or threat actor attribution have been reported (NVIDIA Advisory).

NVIDIA has published Security Bulletin 5865 addressing this and related vulnerabilities in Triton Inference Server. Users should consult the official bulletin at the NVIDIA product-security GitHub repository for patched version details and apply the recommended update. As a general workaround, restricting network access to the Triton Inference Server (e.g., via firewall rules or API gateway rate limiting) can reduce exposure until a patch is applied (NVIDIA Advisory).

The vulnerability received routine automated coverage from vulnerability tracking services such as VulDB, CVEFeed, and INCIBE-CERT shortly after publication. Social media mentions were limited to automated bot posts on Mastodon and Bluesky. No notable independent researcher commentary or significant media coverage has been identified beyond standard aggregator reporting.