Database/AI/ML frameworks & serving

LiteLLM (`/guardrails/test_custom_code`): RCE via bytecode rewriting
CVSS 8.8CVE-2026-40217AI/ML frameworks & servingcurated
Impact
RCE via bytecode rewriting
Who can reach it
Authenticated network user of the proxy
What to do
Upgrade; the guardrail-test endpoint is an eval sink
References
Related entries
- Ray Data (Arrow extension types): Custom Arrow extension types deserialized unsafelyCVE-2026-41486 · Ray Data (Arrow extension types)High
- JupyterLab: extension allow-list not enforced, letting a notebook user install arbitrary PyPI packagesCVE-2026-42266 · JupyterLab (PyPI Extension Manager, allowed_extensions_uris allow-list)High
- LiteLLM proxy: Two endpoints allow privilege escalation / unauthorized actionCVE-2026-42271 · LiteLLM proxyHigh
- BentoML (`bentofile.yaml`): Malicious build manifestCVE-2026-44346 · BentoML (`bentofile.yaml`)High
- diffusers: Hub repo with a None.py file executes code on a plain from_pretrained callCVE-2026-44827 · Hugging Face diffusers (DiffusionPipeline.from_pretrained custom-pipeline resolution)High
- Open WebUI: code execution endpoint runs arbitrary Python even when ENABLE_CODE_EXECUTION is falseCVE-2026-45672 · Open WebUI (code execution endpoint /api/v1/utils/code/execute)High
This entry is curated: imported from vendor advisories with machine assistance, not yet individually verified. Confirm against your vendor's advisory before acting, and report anything wrong.