← Übersicht Nicht im beobachteten Stack

Arbitrary Code Execution via Unsafe torch.load() in Trainer Checkpoint Loading in huggingface/transformers

CVE-2026-1839 · CLSA-2026-1779065183 · GHSA-69w3-r845-3855 · PYSEC-2026-2288

Hoch · CVSS 7.8 Red Hat: Moderate Lokal · Nutzerinteraktion Kein Fix bekannt

Veröffentlicht
07.04.2026 07:22
Zuletzt geändert
17.11.2026
EPSS (30 Tage)
0.4 %
CVSS-Vektor
CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H

Beschreibung

Originaltext der Quelle, nicht übersetzt.

A vulnerability in the HuggingFace Transformers library, specifically in the `Trainer` class, allows for arbitrary code execution. The `_load_rng_state()` method in `src/transformers/trainer.py` at line 3059 calls `torch.load()` without the `weights_only=True` parameter. This issue affects all versions of the library supporting `torch>=2.2` when used with PyTorch versions below 2.6, as the `safe_globals()` context manager provides no protection in these versions. An attacker can exploit this vulnerability by supplying a malicious checkpoint file, such as `rng_state.pth`, which can execute arbitrary code when loaded. The issue is resolved in version v5.0.0rc3.

Status beim Hersteller

Red Hat stuft seine Produktlinien ein: 6 betroffen.

Lightspeed Core BetroffenFix zurückgestellt openshift-lightspeed-tech-preview/lightspeed-rag-tool-rhel9, redhat-user-workloads/lightspeed-core-tenant/lightspeed-stack
OpenShift Lightspeed BetroffenFix zurückgestellt openshift-lightspeed/lightspeed-ocp-rag-rhel9, openshift-lightspeed/lightspeed-service-api-rhel9, redhat-user-workloads/crt-nshift-lightspeed-tenant/lightspeed-rag-tool
Red Hat AI Inference Server BetroffenFix zurückgestellt redhat-user-workloads/ai-tenant/rhaiis-3-2-2/rhaiis-cuda-ubi9-3-2-2, redhat-user-workloads/ai-tenant/rhaiis-3-2-2/rhaiis-model-opt-cuda-ubi9-3-2-2, redhat-user-workloads/ai-tenant/rhaiis-3-2-2/rhaiis-rocm-ubi9-3-2-2, redhat-user-workloads/ai-tenant/rhaiis-3-3/rhaiis-cpu-ubi9-3-3, redhat-user-workloads/ai-tenant/rhaiis-3-3/rhaiis-cuda-ubi9-3-3, redhat-user-workloads/ai-tenant/rhaiis-3-3/rhaiis-model-opt-cuda-ubi9-3-3, redhat-user-workloads/ai-tenant/rhaiis-3-3/rhaiis-neuron-ubi9-3-3, redhat-user-workloads/ai-tenant/rhaiis-3-3/rhaiis-rocm-ubi9-3-3, redhat-user-workloads/ai-tenant/rhaiis-3-3/rhaiis-spyre-ubi9-3-3, redhat-user-workloads/ai-tenant/rhaiis-3-3/rhaiis-tpu-ubi9-3-3, redhat-user-workloads/ai-tenant/rhaiis/rhaiis-spyre-ubi9, redhat-user-workloads/ai-tenant/rhaiis/rhaiis-tpu-ubi9
Red Hat Ansible Automation Platform 2 BetroffenFix zurückgestellt ansible-automation-platform-25/lightspeed-chatbot-rhel8, redhat-user-workloads/ansible-tenant/ansible-automation-platform-26/lightspeed-chatbot-rhel9
Red Hat Enterprise Linux AI (RHEL AI) 3 BetroffenFix zurückgestellt redhat-user-workloads/rhel-ai-tenant/rhelai-bootc-3-3/bootc-cuda-aws-3-3, redhat-user-workloads/rhel-ai-tenant/rhelai-bootc-3-3/bootc-cuda-azure-3-3, redhat-user-workloads/rhel-ai-tenant/rhelai-bootc-3-3/bootc-cuda-gcp-3-3, redhat-user-workloads/rhel-ai-tenant/rhelai-bootc-3-3/bootc-rocm-azure-3-3, rhelai3/bootc-cuda-rhel9, rhelai3/bootc-rocm-rhel9, rhelai3/disk-image-cuda-rhel9
Red Hat OpenShift AI (RHOAI) BetroffenFix zurückgestellt rhoai/odh-built-in-detector-rhel9, rhoai/odh-caikit-nlp-rhel9, rhoai/odh-caikit-tgis-serving-rhel9, rhoai/odh-guardrails-detector-huggingface-runtime-rhel9, rhoai/odh-kserve-agent-rhel9, rhoai/odh-kserve-controller-rhel9, rhoai/odh-kserve-router-rhel9, rhoai/odh-kserve-storage-initializer-rhel9, rhoai/odh-llama-stack-core-rhel9, rhoai/odh-llm-d-inference-scheduler-rhel9, rhoai/odh-openvino-model-server-rhel9, rhoai/odh-pipeline-runtime-pytorch-llmcompressor-cuda-py312-rhel9, rhoai/odh-ta-lmes-job-rhel9, rhoai/odh-training-cuda121-torch24-py311-rhel9, rhoai/odh-training-cuda124-torch25-py311-rhel9, rhoai/odh-training-cuda128-torch28-py312-rhel9, rhoai/odh-training-cuda128-torch29-py312-rhel9, rhoai/odh-training-rocm62-torch24-py311-rhel9, rhoai/odh-training-rocm62-torch25-py311-rhel9, rhoai/odh-training-rocm64-torch28-py312-rhel9, rhoai/odh-training-rocm64-torch29-py312-rhel9, rhoai/odh-trustyai-nemo-guardrails-server-rhel9, rhoai/odh-vllm-cuda-rhel9, rhoai/odh-vllm-gaudi-rhel9, rhoai/odh-vllm-rocm-rhel9 und 2 weitere
Weitere betroffene Pakete (5)
CVEhuggingface/huggingface/transformersv5.0.0rc3CVE-2026-1839
NVDhuggingface:transformers5.0.0CVE-2026-1839
PyPItransformers5.0.0rc3GHSA-69w3-r845-3855
PyPItransformers5.0.0PYSEC-2026-2288
TuxCare:PyPItransformers4.57.6.post0 → 4.57.6.post1+tuxcareCLSA-2026-1779065183

Empfohlene Prüfung

  1. Feststellen, ob und in welcher Version die betroffenen Pakete in Ihren Systemen, Images und Abhängigkeiten vorkommen.
  2. Wo eine Fix-Version genannt ist: aktualisieren. Wo nicht: Hersteller-Advisory und Referenzen auf Workarounds prüfen.
  3. Erreichbarkeit beachten: Der Angriffsweg (Lokal · Nutzerinteraktion) ist aus dem CVSS-Vektor abgeleitet.

Diese Seite sagt nichts darüber, ob Ihre Systeme betroffen sind. Das klärt erst eine Prüfung Ihrer Versionen.