THREATOPS
THREAT OPS › Threat News › [NVD] CVE-2026-115917 (LOW 3.4) — vLLM is an inference and serving engine for large language models. Prior to 0.32.0, the OpenAI-compatible server configures Starlette CORSMiddleware with allowed_origins set to a wildcard by default. A malicious web page can use a victim's browser to reach an unauthenticated vLLM

[NVD] CVE-2026-115917 (LOW 3.4) — vLLM is an inference and serving engine for large language models. Prior to 0.32.0, the OpenAI-compatible server configures Starlette CORSMiddleware with allowed_origins set to a wildcard by default. A malicious web page can use a victim's browser to reach an unauthenticated vLLM

mednvdPublished 2026-10-26

CVE-2026-115917 CVSS: 3.4 LOW Published: 2026-10-26T18:17:39.263

vLLM is an inference and serving engine for large language models. Prior to 0.32.0, the OpenAI-compatible server configures Starlette CORSMiddleware with allowed_origins set to a wildcard by default. A malicious web page can use a victim's browser to reach an unauthenticated vLLM server on localhost, an internal network, or a VPN, c

MITRE ATT&CK techniques

Indicators of compromise

Original source: https://nvd.nist.gov/vuln/detail/CVE-2026-115917