CVE-2026-5817
Medium
Elevated severity or exploit probability.
CVSS base
8.2
HIGH
CVSS:3.1/AV:L/AC:L/PR:L/UI:R/S:C/C:H/I:H/A:H
EPSS — probability of exploitation (30 days)
0.2%
13.4th percentile
CISA KEV
Not listed
Weakness / dates
CWE-829
Published 2026-05-22 · modified 2026-07-23
CVSS breakdown
CVSS:3.1/AV:L/AC:L/PR:L/UI:R/S:C/C:H/I:H/A:H
| Attack Vector | L | Local |
| Attack Complexity | L | Low |
| Privileges Required | L | Low |
| User Interaction | R | Required |
| Scope | C | Changed |
| Confidentiality | H | High |
| Integrity | H | High |
| Availability | H | High |
Timeline
- 2026-05-22 — Published (NVD)
- 2026-07-23 — Last modified (NVD)
Description
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.