AI Component

Inference

Inference servers are the most actively-exploited component of the AI stack because they sit between the model and the public internet and they hold the GPU. The shape of the bugs is mostly web-app classes magnified by the cost of compute: missing auth on /v1 endpoints, SSRF that escapes the sandbox onto the platform's control plane, unsafe deserialization on model-loading paths, and path traversal in artifact-management endpoints. vLLM, Triton, TGI, BentoML, Ray Serve, and Ollama have each shipped multiple high-severity CVEs since 2023; CVE-2024-11041 in vLLM was a notable example combining prompt injection with code execution. Multi-tenant deployments are particularly exposed because a single bug typically crosses tenant boundaries. Defenses: aggressive patching, mandatory auth, network segmentation between inference and control plane, and per-tenant resource quotas to bound abuse.

698
Total CVEs
35
Pages
Page 8 of 35
Current
Severity CVE CVSS
MEDIUM CVE-2021-37688 5.5
MEDIUM CVE-2021-37689 5.5
HIGH CVE-2021-37663 7.8
HIGH CVE-2021-37665 7.8
MEDIUM CVE-2021-37668 5.5
MEDIUM CVE-2021-37669 5.5
MEDIUM CVE-2021-37673 5.5
MEDIUM CVE-2021-37677 5.5
HIGH CVE-2021-37682 7.1
MEDIUM CVE-2021-37683 5.5
MEDIUM CVE-2021-37684 5.5
MEDIUM CVE-2021-37685 5.5
MEDIUM CVE-2021-37687 5.5
MEDIUM CVE-2021-37691 5.5
MEDIUM CVE-2021-41196 5.5
MEDIUM CVE-2021-41197 5.5
MEDIUM CVE-2021-41198 5.5
MEDIUM CVE-2021-41199 5.5
HIGH CVE-2021-41201 7.8
MEDIUM CVE-2021-41204 5.5

Page 8 of 35