CVE-2025-47277: vLLM Remote Code Execution via PyNcclPipe
vLLM
A remote code execution vulnerability affects vLLM environments utilizing the PyNcclPipe KV cache transfer integration. Unauthenticated or improper handling in communication channels permits arbitrary code execution.
disclosed CVENo confirmed exploitation, so this record is tracked as a disclosed vulnerability and not counted as an incident. It moves into the counts if CISA or the vendor confirms exploitation. Why
- CVSS
- 9.8
- EPSS
- 0.01 · top 38% · 2026-11-13
- fix status
- Fix status unknown
- first report
- 20 May 2025
- last update
- 20 May 2025
- CVE ids
Timeline · 2
Sources · 2
Advisories and databases
- 2025-05-20CVE-2025-47277: vLLM Allows Remote Code Execution via PyNcclPipe Communication ServiceGitHub Advisory Database · github.com/advisories/GHSA-hjq4-87xh-g4fv
- 2025-05-20CVE-2025-47277: vLLM, an inference and serving engine for large language models (LLMs), has an issue in versions 0.6.5 through 0.8.4 that ONLY…NVD · nvd.nist.gov/vuln/detail/CVE-2025-47277
Titles and links belong to their publishers. The summary above is written by gcdRadar in its own words; data is CC BY-SA 4.0.
Classification
- lane
- Patch now
- AI role
- Attack on AI
- categories
- Infrastructure vulnerability
- severity
- 2 · Disclosed vulnerability
- fix status
- Fix status unknown
- targets
- Framework / tooling
- owasp llm top 10
- LLM03:2025 Supply Chain
- mitre atlas
- no technique for these categories
- dates
- happened 20 May 2025 · first reported 20 May 2025
- origin
- GitHub Advisory DatabaseVulnerability record from NVD / the GitHub Advisory Database. This product uses the NVD API but is not endorsed or certified by the NVD.
- record
- confidence 0.95 · created 2026-10-18 07:00Z · updated 2026-10-18 10:04Z
Lane, AI role, categories and severity come from the classifier; the OWASP and ATLAS rows are derived from the categories. How these are assigned
Copy a brief
This record as a short Markdown brief, for a ticket, a chat or a coding agent.
the brief as text
# CVE-2025-47277: vLLM Remote Code Execution via PyNcclPipe https://gcdradar.vercel.app/incidents/cve-2025-47277-vllm-remote-code-execution-via-pyncclpipe · AIR-2025-0285 · closed 2025-05-20 - Lane: Patch now - Severity: 2 (Disclosed vulnerability) - Fix status: Fix status unknown - First reported: 2025-05-20 - AI role: Attack on AI - Products: vLLM - CVE ids: CVE-2025-47277 - CVSS: 9.8 ## Summary A remote code execution vulnerability affects vLLM environments utilizing the PyNcclPipe KV cache transfer integration. Unauthenticated or improper handling in communication channels permits arbitrary code execution. ## What to do No action recorded; read the sources below and the vendor's advisory. ## Latest updates - 2025-05-20: CVE-2025-47277: vLLM, an inference and serving engine for large language models (LLMs), has an issue in versions 0.6.5 through 0.8.4 that ONLY… (https://nvd.nist.gov/vuln/detail/CVE-2025-47277) - 2025-05-20: CVE-2025-47277: vLLM Allows Remote Code Execution via PyNcclPipe Communication Service (https://github.com/advisories/GHSA-hjq4-87xh-g4fv) ## Sources - 2025-05-20 GitHub Advisory Database: CVE-2025-47277: vLLM Allows Remote Code Execution via PyNcclPipe Communication Service https://github.com/advisories/GHSA-hjq4-87xh-g4fv - 2025-05-20 NVD: CVE-2025-47277: vLLM, an inference and serving engine for large language models (LLMs), has an issue in versions 0.6.5 through 0.8.4 that ONLY… https://nvd.nist.gov/vuln/detail/CVE-2025-47277 Verify against the vendor's advisory before changing production. Data CC BY-SA 4.0, https://gcdradar.vercel.app.
Related
- closed6.5vLLM CVE-2025-48943: vLLM Denial of Service via Invalid Regex Output
Patch now · Disclosed · since 28 May 2025 · updated 30 May 2025 · CVE-2025-48943
AI - closed6.5vLLM CVE-2025-48942: vLLM Denial of Service via Invalid JSON Schema
Patch now · Disclosed · since 28 May 2025 · updated 30 May 2025 · CVE-2025-48942
AI - closed6.5vLLM CVE-2025-48887: vLLM ReDoS Vulnerability in pythonic_tool_parser
Patch now · Disclosed · since 28 May 2025 · updated 30 May 2025 · CVE-2025-48887
AI - closedvLLM vLLM Regular Expression Denial of Service
Patch now · Disclosed · since 28 May 2025
AI - closed4.2vLLM CVE-2025-46722: vLLM MultiModalHasher Image Hashing Weakness
Patch now · Disclosed · since 28 May 2025 · updated 29 May 2025 · CVE-2025-46722
AI - closed2.6vLLM CVE-2025-46570: Timing Side-Channel in vLLM Prefix Caching
Patch now · Disclosed · since 28 May 2025 · updated 29 May 2025 · CVE-2025-46570
AI