Incomplete artifact pin propagation in FunAudioChat and Tarsier2
GHSA-hhv2-872h-628q · published 2026-09-12
- Severity
- Medium
- CVSS score
- 6.5
- Chance of exploitation (EPSS)
- No CVE
- Fixed in
- 0.28.0
Operators who pin FunAudioChat or Tarsier2 deployments to a reviewed Hugging Face revision can still load behavior-affecting processor/tokenizer/config artifacts from the repository default revision. A later default-branch change can alter audio preprocessing, speech tokenizer behavior, or Tarsier2 config without changing the operator's configured revision. This is a supply-chain integrity and reproducibility failure for pinned deployments. It does not claim unauthenticated RCE or a trust_remote_code=False bypass.
Upgrade vllm to 0.28.0 or later. If you start it from an MCP config with npx or uvx, pin that version so an old cached copy is not used.
Forward the configured revision at the remaining boundaries. The speech tokenizer should use model_config.tokenizer_revision because vLLM exposes that separate pin and normalizes it to revision when unset. Suggested severity: Moderate. Suggested CVSS: (6.5). Suggested CWE: CWE-348, with CWE-345 also reasonable for consistency with GHSA-3ww4-5jv9-j5gm.
| Package | Registry | Affected | Fixed in |
|---|---|---|---|
| vllm | PyPI | >= 0.22.1, < 0.28.0 | 0.28.0 |
- CriticalLiteLLM has SQL Injection in Proxy API key verificationlitellmExploited
- Criticaln8n Vulnerable to Remote Code Execution via Expression Injectionn8n, n8nExploited
- HighLiteLLM has a sandbox escape in custom-code guardraillitellmExploited
- HighLangflow affected by Remote Code Execution via validate_code() exec()langflowExploited
- CriticalJupyter MCP Server: Browser DNS rebinding can hijack the Jupyter backend through management routesjupyter-mcp-server
- Criticalmem0ai Databricks vector store vulnerable to SQL injectionmem0ai
- CriticalvLLM: OpenAI auth bypassvllm
- CriticalvLLM has RCE In Video Processingvllm
- CriticalvLLM Allows Remote Code Execution via PyNcclPipe Communication Servicevllm
- CriticalvLLM Vulnerable to Remote Code Execution via Mooncake Integrationvllm
- CriticalCVE-2025-24357 Malicious model remote code execution fix bypass with PyTorch < 2.6.0vllm
- CriticalvLLM Deserialization of Untrusted Data vulnerabilityvllm
The impact, patch and workaround text comes from the advisory, shortened (published by vLLM). Proof-of-concept and reproduction sections and code are left out; the original advisory has them.