Snyk has a proof-of-concept or detailed explanation of how to exploit this vulnerability.
The probability is the direct output of the EPSS model, and conveys an overall sense of the threat of exploitation in the wild. The percentile measures the EPSS probability relative to all known EPSS scores. Note: This data is updated daily, relying on the latest available EPSS model version. Check out the EPSS documentation for more details.
In a few clicks we can analyze your entire application and see what components are vulnerable in your application, and suggest you quick fixes.
Test your applicationsUpgrade vllm to version 0.29.0 or higher.
vllm is an A high-throughput and memory-efficient inference and serving engine for LLMs
Affected versions of this package are vulnerable to Improper Validation of Specified Quantity in Input via load_audio_pyav() in the audio transcription endpoint (/v1/audio/transcriptions). The function applies only a compressed file-size check before passing uploaded audio bytes to av.open(), allowing a malformed IAMF (Immersive Audio Model and Formats) container to reach FFmpeg's native libavformat IAMF parser unvalidated. The parser's scalable_channel_layout_config() function performs a write past a 32-byte AVChannelCustom heap allocation, corrupting the allocator heap. Repeated uploads accumulate corruption until the API process crashes, denying service to all users of that server.
Note: This is only exploitable when the deployment serves the speech endpoint, soundfile is unavailable or cannot parse the container causing fallback to PyAV, and the linked libavformat is within the vulnerable IAMF parser range (confirmed on PyAV 16.1.0 / FFmpeg 8.0.1).