Threat Advisory

vLLM Flaw Risks Remote Code Execution via Malicious Prompt Embeddings

Threat: Vulnerability
Targeted Region: Global
Targeted Sector: Technology & IT
Criticality: High
[subscribe_to_unlock_form]

EXECUTIVE SUMMARY:

A high-severity CVE-2025-62164, CVSS 8.8 vulnerability in the vLLM inference engine versions 0.10.2 to before 0.11.1 allows attackers to submit malicious prompt embeddings to the Completions API, triggering a memory-corruption bug (due to unsafe deserialization via torch.load) that can lead to denial-of-service or even remote code execution.[/subscribe_to_unlock_form]

EXECUTIVE SUMMARY:

A high-severity CVE-2025-62164, CVSS 8.8 vulnerability in the vLLM inference engine versions 0.10.2 to before 0.11.1 allows attackers to submit malicious prompt embeddings to the Completions API, triggering a memory-corruption bug (due to unsafe deserialization via torch.load) that can lead to denial-of-service or even remote code execution.[emaillocker id="1283"]

RECOMMENDATION:

We strongly recommend you update vLLM to version 0.11.1.

REFERENCES:

The following reports contain further technical details:

https://securityonline.info/vllm-flaw-cve-2025-62164-risks-remote-code-execution-via-malicious-prompt-embeddings/

[/emaillocker]
crossmenu