CVE-2026-27893: vLLM's hardcoded trust_remote_code=True in NemotronVL and KimiK25 bypasses user security opt-out

Severity: high (CVSS 8.8 v3.1)
Exploited in the wild: not listed in CISA KEV
EPSS: 1.4% probability of exploitation in the next 30 days
Public exploit or detection template: none known
Published: 2026-03-27 · updated 2026-07-21
Weakness: CWE-693

Description

vLLM is an inference and serving engine for large language models (LLMs). Starting in version 0.10.1 and prior to version 0.18.0, two model implementation files hardcode `trust_remote_code=True` when loading sub-components, bypassing the user's explicit `--trust-remote-code=False` security opt-out. This enables remote code execution via malicious model repositories even when the user has explicitly disabled remote code trust. Version 0.18.0 patches the issue.

Affected products

vllm-project vllm: from 0.10.1 before 0.18.0 — fixed in 0.18.0

How to fix

Upgrade to a fixed version: vllm-project vllm 0.18.0. Check the vendor advisory for mitigations.

References

Get alerted when CVE-2026-27893 or the next vulnerability affects your stack