← Vulnerability feed

Vulnerability record · CVE-2026-80047 · published 1 September 2026

CVE-2026-80047: TOCTOU race condition vulnerability

A vulnerability in Hugging Face Transformers (versions 4.57.0 to 5.16.1) allows remote Python files to be written to local disk without user consent when using GenerativePreTrainedModel.load_custom_generate(). The function fetches and caches a remote module file before performing the required trust_remote_code consent check, inverting the security model enforced by other code-loading paths (such as AutoConfig, AutoModel, and AutoTokenizer). As a result, attacker‑controlled Python code from custom_generate/generate.py is copied into the user’s ~/.cache/huggingface/modules directory even if the user declines the trust prompt. Although execution is correctly gated, the file write is not reversible and can persist across sessions. This can lead to persistent, unauthorized files on disk and stale cache collisions where cached attacker code may later be executed during trusted model loads. The issue stems from an unconditional file write in dynamic_module_utils.py prior to any trust verification.

7.8 CVSS 3.1 High EPSS 0.10% · top 99.2% CWE-273 · CWE-273CWE-367 · TOCTOU race condition Awaiting Analysis
7.8CVSS 3.1 base score
0.10%EPSS exploitation probability, 30 days
NoNot in CISA KEV
0Affected product versions listed by NVD
3References
23 Sep 2026Last modified by NVD

Description

A vulnerability in Hugging Face Transformers (versions 4.57.0 to 5.16.1) allows remote Python files to be written to local disk without user consent when using GenerativePreTrainedModel.load_custom_generate(). The function fetches and caches a remote module file before performing the required trust_remote_code consent check, inverting the security model enforced by other code-loading paths (such as AutoConfig, AutoModel, and AutoTokenizer). As a result, attacker‑controlled Python code from custom_generate/generate.py is copied into the user’s ~/.cache/huggingface/modules directory even if the user declines the trust prompt. Although execution is correctly gated, the file write is not reversible and can persist across sessions. This can lead to persistent, unauthorized files on disk and stale cache collisions where cached attacker code may later be executed during trusted model loads. The issue stems from an unconditional file write in dynamic_module_utils.py prior to any trust verification.

CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H

References

LinkTags
https://github.com/huggingface/transformers
https://kb.cert.org/vuls/id/456290
https://www.kb.cert.org/vuls/id/456290

Track CVE-2026-80047 inside VULONE

Watch it alongside the ransomware crews, C2 infrastructure and forum chatter that reference it, query it through the API and pull it into your SIEM over TAXII.

Start free Open in platform

Related vulnerabilities

Same products first, then exploited flaws of the same weakness class.

9.8CVE-2024-38813VMware vCenter Server privilege escalation to root via crafted packetvCenter Server contains a privilege escalation flaw where a malicious actor with network access can send a specially crafted network packet to escala…KEVEPSS 17%analysed

Source: NIST National Vulnerability Database (record CVE-2026-80047), CISA KEV, FIRST EPSS (scores of 2026-09-26). This page is refreshed as NVD updates the record.