{"uuid": "e633c16f-9296-4d26-ba3b-d0be1f631aa6", "vulnerability_lookup_origin": "1a89b78e-f703-45f3-bb86-59eb712668bd", "author": "9f56dd64-161d-43a6-b9c3-555944290a09", "vulnerability": "CVE-2026-57173", "type": "seen", "source": "https://gist.github.com/alon710/725c0a2c00f7edc1e74698f411f27baa", "content": "# CVE-2026-57173: CVE-2026-57173: Unauthenticated Audio Decompression-Bomb Denial of Service in vLLM\n\n> **CVSS Score:** 6.5\n> **Published:** 2026-09-16\n> **Full Report:** https://cvereports.com/reports/CVE-2026-57173\n\n## Summary\nCVE-2026-57173 (GHSA-hcwq-8wjf-3gcr) represents a critical resource allocation validation vulnerability in the vLLM inference engine. Prior to version 0.24.0, vLLM's multimodal chat completions pipeline failed to enforce maximum audio decode duration limits. Unauthenticated remote attackers can exploit this to perform an audio decompression bomb attack, causing massive memory allocations that trigger immediate system Out-Of-Memory (OOM) crashes and service termination.\n\n## TL;DR\nvLLM versions prior to 0.24.0 are vulnerable to remote Denial of Service via an audio decompression bomb targeting the chat completions API, causing an immediate worker crash due to memory exhaustion.\n\n## Exploit Status: POC\n\n## Technical Details\n\n- **CWE ID**: CWE-770\n- **Attack Vector**: Network\n- **CVSS v3.1 Score**: 6.5\n- **EPSS Score**: Not Recorded\n- **Impact**: Endpoint Denial of Service (OOM Crash)\n- **Exploit Status**: Proof of Concept (PoC)\n- **KEV Status**: Not Listed\n\n## Affected Systems\n\n- vLLM installations serving audio-capable multimodal models\n- **vLLM**: < 0.24.0 (Fixed in: `0.24.0`)\n\n## Mitigation\n\n- Upgrade vLLM to version 0.24.0 or higher.\n- Limit the maximum request size on proxy or ingress layers to block large inline payload delivery.\n- Configure the VLLM_MAX_AUDIO_DECODE_DURATION_S environment variable to a minimal acceptable value.\n- Deploy API-level authentication to prevent external unauthenticated access to serving endpoints.\n\n**Remediation Steps:**\n1. Step 1: Terminate current vulnerable instances of vLLM.\n2. Step 2: Update the application environment to vLLM >= 0.24.0.\n3. Step 3: Define VLLM_MAX_AUDIO_DECODE_DURATION_S=300 in the runner's execution environment.\n4. Step 4: Restart the vLLM serving container and verify that API requests are processed within safety limits.\n\n## References\n\n- [GHSA-hcwq-8wjf-3gcr](https://github.com/vllm-project/vllm/security/advisories/GHSA-hcwq-8wjf-3gcr)\n- [Fix Commit](https://github.com/vllm-project/vllm/commit/3d20275bb4d434f53055c3c0b645fd8bb072965e)\n\n\n---\n*Generated by [CVEReports](https://cvereports.com/reports/CVE-2026-57173) - Automated Vulnerability Intelligence*", "creation_timestamp": "2026-09-17T05:32:04.684548Z"}