Transformers save_pretrained path traversal allows arbitrary file writes through chat template names
PYSEC-2026-3929 · CVE-2026-9856 · GHSA-xrqw-3rrv-vx5w
Published · Modified
AI SAST
Find this class of vulnerability in your own code
Corgea's AI-native static analysis detects vulnerabilities like this one across your repositories, ranks them by exploitability, and returns review-ready fixes.
Description
A vulnerability in huggingface/transformers versions < 5.10.0 allows an attacker to perform arbitrary file writes via path traversal. The issue resides in the save_pretrained() methods of PreTrainedTokenizerBase and ProcessorMixin, where keys from the chat_template dictionary are used directly as filenames without proper validation. An attacker can exploit this by publishing a malicious Hugging Face Hub repository with a crafted tokenizer_config.json file. When a victim downloads and saves the tokenizer or processor, the attacker-controlled keys can escape the intended save directory, enabling arbitrary file writes with attacker-controlled content. This vulnerability affects multiple processors inheriting from ProcessorMixin, including Idefics, Florence, Gemma, Phi, and Qwen-VL.
References
- ADVISORY https://nvd.nist.gov/vuln/detail/CVE-2026-9856
- WEB https://github.com/huggingface/transformers/pull/46191
- WEB https://github.com/huggingface/transformers/commit/eaaaf8494dd5386634ae37d1d122212fdc315be5
- PACKAGE https://github.com/huggingface/transformers
- WEB https://huntr.com/bounties/362824d5-fe18-40e8-a6cf-62277f97a170
- PACKAGE https://pypi.org/project/transformers
- ADVISORY https://github.com/advisories/GHSA-xrqw-3rrv-vx5w
Ready to move
Start Securing
Free, no credit card | First findings in minutes