PT-2026-42830 · Docker · Docker Desktop

David Rochester

+1

·

Published

2026-05-22

·

Updated

2026-05-22

·

CVE-2026-5817

CVSS v3.1

8.2

High

VectorAV:L/AC:L/PR:L/UI:R/S:C/C:H/I:H/A:H
The vllm-metal inference backend in Docker Model Runner on macOS unconditionally sets trust remote code=True when loading model tokenizers, and runs without sandboxing. This causes transformers.AutoTokenizer.from pretrained() to import and execute arbitrary Python files included in any model pulled from an OCI registry, resulting in arbitrary code execution on the Docker host as the Docker Desktop user when inference is triggered.
Any container on the Docker network can trigger this by calling the model-runner.docker.internal API to pull a malicious model and request inference.

Fix

Weakness Enumeration

Related Identifiers

CVE-2026-5817

Affected Products

Docker Desktop