Description
Axolotl before 0.19.0 contains a remote code execution vulnerability in the multipack patch path where trust_remote_code defaults to None instead of False, causing the security guard to be bypassed. Attackers can execute arbitrary Python code by crafting a malicious Hugging Face model repository selected as base_model, which is loaded with hardcoded trust_remote_code=True during AutoModelForCausalLM.from_pretrained.
Affected products
| Vendor | Product | Affected | Fixed |
|---|---|---|---|
| axolotl-ai-cloud | axolotl | < 0.19.0 | 0.19.0 |
References
- Pull Request #3858 patch issue-tracking
- Patch Commit patch
- Multipack trust_remote_code guard technical-description
- patch_remote sink technical-description
- axolotl v0.19.0 Release Notes release-notes
- https://github.com/axolotl-ai-cloud/axolotl product
- VulnCheck Advisory: Axolotl through 0.18.0 Remote Code Execution via Multipack Patching third-party-advisory