CVE-2026-90555

vLLM versions before 0.28.0 fail to validate audio sample rate headers in the transcription endpoint, allowing authenticated clients to bypass duration checks. Attackers can submit forged FLAC headers with inflated sample rates to trigger excessive memory allocation and crash the API server proce...

CVE-2026-90555

A flaw was found in vLLM. An authenticated client can exploit this vulnerability by submitting forged FLAC Free Lossless Audio Codec headers with an inflated sample rate to the transcription endpoint. This bypasses duration checks, leading to excessive memory allocation within the API server...

CVE-2026-90555

vLLM versions before 0.28.0 contain a Denial of Service vulnerability in the transcription endpoint. The root cause is a failure to validate audio sample rate headers in submitted files. An authenticated attacker can bypass duration checks by submitting forged FLAC headers with inflated sample ra...

EUVD-2026-76873

vLLM versions before 0.28.0 fail to validate audio sample rate headers in the transcription endpoint, allowing authenticated clients to bypass duration checks. Attackers can submit forged FLAC headers with inflated sample rates to trigger excessive memory allocation and crash the API server proce...

PT-2026-90660

vLLM versions before 0.28.0 fail to validate audio sample rate headers in the transcription endpoint, allowing authenticated clients to bypass duration checks. Attackers can submit forged FLAC headers with inflated sample rates to trigger excessive memory allocation and crash the API server proce...

CVE-2026-18258

CVE-2026-18258 affects Scripta/eScriptorium up to version 26.04.1, impacting the Line, LineTranscription, VirtualCollection, tag and process API endpoints. The root cause is authorization bypass where a remote authenticated user can read, modify, and delete other users’ transcription content. Thi...

GHSA-V82G-2437-67M2 vLLM: Speech-to-text upload size limit is enforced after full UploadFile read

Summary Current-head vLLM documents VLLMMAXAUDIOCLIPFILESIZEMB as the maximum audio file size accepted by the speech-to-text APIs. The default is 25 MB. vllm/envs.py also describes files larger than this value as rejected. The /v1/audio/transcriptions and /v1/audio/translations routes call await...

PYSEC-2026-3404 vLLM: OOM Denial of Service via Audio Decompression Bomb

Summary vLLM's /v1/audio/transcriptions endpoint limits compressed upload size but not decoded PCM output. A 25MB OPUS file expands to 14.9GB of float32 PCM at decode time. Tested on vLLM v0.19.0. Details SpeechToTextProcessor rejects uploads over VLLMMAXAUDIOCLIPFILESIZEMB default 25MB based on...

PYSEC-2026-2764 Open WebUI vulnerable to Path Traversal in `POST /api/v1/audio/transcriptions`

Summary An unsanitised filename field in the speech-to-text transcription endpoint allows any authenticated non-admin user to trigger a FileNotFoundError whose message — including the server's absolute DATADIR path — is returned verbatim in the HTTP 400 response body, confirming information...

CVE-2026-55646 vLLM speech-to-text endpoints allocate full upload before enforcing the audio file-size limit

vLLM is an inference and serving engine for large language models. From 0.22.0 to 0.23.0, the /v1/audio/transcriptions and /v1/audio/translations routes call request.file.read to fully materialize an uploaded audio file into memory before vLLM checks the documented VLLMMAXAUDIOCLIPFILESIZEMB...

CVE-2026-54233

A flaw was found in vLLM, an inference and serving engine for large language models LLMs. A remote attacker could exploit a vulnerability in the /v1/audio/transcriptions endpoint. By uploading a specially crafted compressed audio file, such as an OPUS file, the attacker could cause the system to...

CVE-2026-54233 vLLM: OOM Denial of Service via Audio Decompression Bomb

vLLM is an inference and serving engine for large language models LLMs. Prior to 0.23.1rc0, vLLM's /v1/audio/transcriptions endpoint limits compressed upload size but not decoded PCM output. A 25MB OPUS file expands to 14.9GB of float32 PCM at decode time. This vulnerability is fixed in 0.23.1rc0...

vLLM: OOM Denial of Service via Audio Decompression Bomb

Summary vLLM's /v1/audio/transcriptions endpoint limits compressed upload size but not decoded PCM output. A 25MB OPUS file expands to 14.9GB of float32 PCM at decode time. Tested on vLLM v0.19.0. Details SpeechToTextProcessor rejects uploads over VLLMMAXAUDIOCLIPFILESIZEMB default 25MB based on...

GHSA-6PR9-RP53-2PMC vLLM: OOM Denial of Service via Audio Decompression Bomb

Summary vLLM's /v1/audio/transcriptions endpoint limits compressed upload size but not decoded PCM output. A 25MB OPUS file expands to 14.9GB of float32 PCM at decode time. Tested on vLLM v0.19.0. Details SpeechToTextProcessor rejects uploads over VLLMMAXAUDIOCLIPFILESIZEMB default 25MB based on...

PT-2026-50489

Name of the Vulnerable Software and Affected Versions vLLM versions prior to 0.23.1rc0 Description The '/v1/audio/transcriptions' endpoint limits the size of compressed uploads but fails to limit the size of the decoded PCM Pulse Code Modulation output. PCM is an uncompressed digital audio format...

CVE-2026-25058 Vexa's unauthenticated internal transcript endpoint exposed by default

Vexa is an open-source, self-hostable meeting bot API and meeting transcription API. Prior to 0.10.0-260419-1910, the Vexa transcription-collector service exposes an internal endpoint GET /internal/transcripts/meetingid that returns transcript data for any meeting without any authentication or...

Information Exposure

Overview open-webui is an Open WebUI Affected versions of this package are vulnerable to Information Exposure via the POST /api/v1/audio/transcriptions endpoint. An attacker can obtain sensitive server filesystem path information by submitting a crafted multipart request with a malicious filename...

GHSA-VVXM-VXMR-624H Open WebUI vulnerable to Path Traversal in `POST /api/v1/audio/transcriptions`

Summary An unsanitised filename field in the speech-to-text transcription endpoint allows any authenticated non-admin user to trigger a FileNotFoundError whose message — including the server's absolute DATADIR path — is returned verbatim in the HTTP 400 response body, confirming information...

CVE-2026-28786 Open WebUI vulnerable to Path Traversal in `POST /api/v1/audio/transcriptions`

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to version 0.8.6, an unsanitized filename field in the speech-to-text transcription endpoint allows any authenticated non-admin user to trigger a FileNotFoundError whose message — including th...

CVE-2026-28786 Open WebUI vulnerable to Path Traversal in `POST /api/v1/audio/transcriptions`

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to version 0.8.6, an unsanitized filename field in the speech-to-text transcription endpoint allows any authenticated non-admin user to trigger a FileNotFoundError whose message — including th...