要点
- CVE-2026-54232 は vllm, vllm (vllm) に影響する深刻度「高」 (CVSS 8.8) の脆弱性です。
- 今後30日間に悪用される確率（EPSS）：0.56%（全脆弱性の 46.8% より高い）。
- 対処：0.22.1 以降に更新してください。
| パッケージ | 影響を受けるバージョン | 修正済みバージョン |
|---|
| vllm PyPI | < 0.22.1
| 0.22.1
|
| vllm | < 0.22.1
| 0.22.1
|
概要
vLLM is an inference and serving engine for large language models (LLMs). Prior to 0.22.1, the vLLM Dockerfile is vulnerable to a dependency confusion attack through the flashinfer-jit-cache package. The package is installed from a custom index (flashinfer.ai/whl/) using --extra-index-url, but the package name was not registered on PyPI, and UV_INDEX_STRATEGY="unsafe-best-match" is set globally. An attacker who registers flashinfer-jit-cache on PyPI with version 0.6.11.post2 can execute arbitrary code as root during the Docker build and backdoor every resulting container image, enabling exfiltration of all user prompts, API credentials, and model data from production vLLM deployments This vulnerability is fixed in 0.22.1.
NVD の説明
vLLM is an inference and serving engine for large language models (LLMs). Prior to 0.22.1, the vLLM Dockerfile is vulnerable to a dependency confusion attack through the flashinfer-jit-cache package. The package is installed from a custom index (flashinfer.ai/whl/) using --extra-index-url, but the package name was not registered on PyPI, and UV_INDEX_STRATEGY="unsafe-best-match" is set globally. An attacker who registers flashinfer-jit-cache on PyPI with version 0.6.11.post2 can execute arbitrary code as root during the Docker build and backdoor every resulting container image, enabling exfiltration of all user prompts, API credentials, and model data from production vLLM deployments This vulnerability is fixed in 0.22.1.
これらの製品をお使いですか？サイトをスキャンして、脆弱なバージョンが公開されていないか確認しましょう。
無料でサイトをスキャン