{
 "id": "AIR-2025-0327",
 "slug": "cve-2025-48956-vllm-api-denial-of-service-via-large-http-hea",
 "title": "CVE-2025-48956: vLLM API Denial of Service via Large HTTP Headers",
 "summary": "An unauthenticated denial of service vulnerability in vLLM occurs when processing excessively large HTTP request headers, leading to server memory exhaustion. Vendor updates are available to resolve the problem.",
 "type": "attack_on_ai",
 "categories": [
  "infra_vuln"
 ],
 "severity": 2,
 "targets": [
  "framework_or_tooling"
 ],
 "products": [
  "vLLM"
 ],
 "incident_date": "2025-08-21",
 "incident_date_precision": "day",
 "first_reported": "2025-08-21",
 "sources": [
  {
   "url": "https://github.com/advisories/GHSA-rxc4-3w6r-4v47",
   "publisher": "GitHub Advisory Database",
   "title": "CVE-2025-48956: vllm API endpoints vulnerable to Denial of Service Attacks",
   "published_at": "2025-08-21"
  },
  {
   "url": "https://nvd.nist.gov/vuln/detail/CVE-2025-48956",
   "publisher": "NVD",
   "title": "CVE-2025-48956: vLLM is an inference and serving engine for large language models (LLMs).",
   "published_at": "2025-08-21"
  }
 ],
 "source_count": 2,
 "confidence": 0.95,
 "status": "published",
 "origin": "seed_ghsa",
 "tags": [
  "CVE-2025-48956",
  "dos"
 ],
 "created_at": "2026-10-18T07:00:14.101Z",
 "updated_at": "2026-10-18T10:04:21.885Z"
}