CVE-2026-71281

8.8 HIGHpublic exploit available

Published 2026-08-05 · Updated 2026-08-10

AI risk analysis

Summary: The flaw allows attackers to execute arbitrary code by providing malicious cache/covariance files, as the torch.load function is called without weights_only=True, bypassing safety checks.
Exploitability: Exploitation requires control over config-specified cache/covariance files and knowledge of file paths. Difficulty varies based on access controls.
Blast radius: If exploited, this could lead to full system compromise or data corruption affecting the Hugging Face peft library users.
Prioritized remediation: Update to the latest version of the affected libraries that include proper safety checks.

rcecode-executiontorchsecurity

Analysis generated locally by qwen2.5:7b-instruct (no data left the box). AI-assisted — verify against primary sources before acting.

NVD description

Hugging Face peft's LoRA-GA and CorDA initialization modules (src/peft/tuners/lora/corda.py lines ~102 and ~163, and src/peft/tuners/lora/loraga.py line ~101) call torch.load on config-specified cache/covariance files without weights_only=True, bypassing peft's own safe-loading wrapper used elsewhere in the codebase.

CVSS vector

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

Weaknesses

CWE-502

Public exploit & PoC references

All references

Source data: NVD (nvd.nist.gov), public domain. Exploit-DB.ai adds local AI analysis for defensive use only.