Vulnerability · Published May 28, 2026 · Assigned by @huntr_ai
CVE-2026-4944Hardcoded trust_remote_code=True in vllm-project/vllm Bypasses User Security Control
High 8.8
In plain English
A high-severity path traversal in vllm-project/vllm from vllm-project. Proof-of-concept exploit code is public. The chance it is exploited in the next 30 days is estimated at 0.7% (higher than 55% of all known vulnerabilities).
Official description
vllm-project/vllm version 0.14.1 contains a vulnerability where the `trust_remote_code=True` parameter is hardcoded in two model implementation files (`vllm/model_executor/models/nemotron_vl.py` and `vllm/model_executor/models/kimi_k25.py`). This bypasses the user's explicit `--trust-remote-code=False` setting, enabling remote code execution via malicious HuggingFace model repositories. This issue is an incomplete fix for CVE-2025-66448 and CVE-2026-22807, as it affects separate code paths in model implementation files. Deployments loading NemotronVL or KimiK25 models are particularly impacted.
Affected products (as reported)
| Vendor | Product | Versions |
|---|---|---|
| vllm-project | vllm-project/vllm | unspecified – through latest |
Weakness: CWE-22 · Vector: CVSS:3.0/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H