CVE-2026-5817
highpriority · track · 44EPSS 0.002 · 13 %ileexploit · PoCDEEURUUSEPSS · scored 2026-10-27
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.
cvss
Technical severity, 0–10. One row per scoring source.
base score 0–10 · higher = more severe · hover a vector to decode
ssvc
A stakeholder decision — act, attend, or track — from exploitation and impact.
weaknesses
The CWE weakness classes this flaw belongs to.
exploits (2)
Public proof-of-concept and weaponised exploit code.
affected products (5)
Products and version ranges known to be affected.
advisories (1)
Vendor and ecosystem advisories describing this.
- WID-SEC-W-2026-1662Docker Desktop Model Runner: Mehrere Schwachstellen ermöglichen CodeausführungDeutsch2026-05-25csaf · cert_bundDE
Ein Angreifer kann mehrere Schwachstellen in Docker Desktop Model Runner ausnutzen, um beliebigen Programmcode auszuführen.
also known as (4)
Other names and identifiers for this entity.
sources
Which sources assert this record, and when.
references (4)
External citations — patches, advisories, analysis.
Attackers weaponise vulnerabilities like this against internet-exposed services. See what is exposed on your perimeter with a free passive scan.
Free scan