{
 "id": "AIR-2025-0197",
 "slug": "cve-2025-25183-vllm-vulnerable-to-hash-collisions-in-prefix",
 "title": "CVE-2025-25183: vLLM vulnerable to hash collisions in prefix cache",
 "summary": "An issue in vLLM allows maliciously constructed prompts to cause hash collisions in the prefix cache, leading to cache reuse and unintended behavior. Patches or mitigations are recommended for affected deployments.",
 "type": "attack_on_ai",
 "categories": [
  "infra_vuln"
 ],
 "severity": 2,
 "targets": [
  "framework_or_tooling"
 ],
 "products": [
  "vLLM"
 ],
 "incident_date": "2025-02-06",
 "incident_date_precision": "day",
 "first_reported": "2025-02-06",
 "sources": [
  {
   "url": "https://github.com/advisories/GHSA-rm76-4mrf-v9r8",
   "publisher": "GitHub Advisory Database",
   "title": "CVE-2025-25183: vLLM uses Python 3.12 built-in hash() which leads to predictable hash collisions in prefix cache",
   "published_at": "2025-02-06"
  },
  {
   "url": "https://nvd.nist.gov/vuln/detail/CVE-2025-25183",
   "publisher": "NVD",
   "title": "CVE-2025-25183: vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs.",
   "published_at": "2025-02-07"
  }
 ],
 "source_count": 2,
 "confidence": 0.95,
 "status": "published",
 "origin": "seed_ghsa",
 "tags": [
  "CVE-2025-25183",
  "vllm",
  "cache-collision"
 ],
 "created_at": "2026-10-18T07:00:14.101Z",
 "updated_at": "2026-10-18T10:04:21.885Z"
}