Vulnerability Description
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, call endpoints including GET /v1/models and POST /v1/chat/completions, consume GPU resources, and read responses. Cookie-based cross-site request forgery does not apply because credentials are disabled by default, and deployments requiring bearer authentication are exposed only when the caller already has a usable token. This issue is fixed in version 0.32.0.
CVSS Score
LOW
Affected Products
| Vendor | Product | Versions |
|---|---|---|
| Vllm | Vllm | < 0.32.0 |
Related Weaknesses (CWE)
References
- https://github.com/vllm-project/vllm/commit/fcae5c98870ff2c20659731dce7ba3a3ff0dPatch
- https://github.com/vllm-project/vllm/pull/61009Patch
- https://github.com/vllm-project/vllm/releases/tag/v0.32.0Release Notes
- https://github.com/vllm-project/vllm/security/advisories/GHSA-vvpg-rrww-wpmhPatchVendor Advisory
FAQ
What is CVE-2026-115917?
CVE-2026-115917 is a vulnerability with a CVSS score of 3.4 (LOW). 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. ...
How severe is CVE-2026-115917?
CVE-2026-115917 has been rated LOW with a CVSS base score of 3.4/10. Review the CVSS metrics above for detailed severity breakdown.
Is there a patch for CVE-2026-115917?
Check the references section above for vendor advisories and patch information. Affected products include: Vllm Vllm.