CVE-2026-6859
| Summary |
A flaw was found in InstructLab. The `linux_train.py` script hardcodes `trust_remote_code=True` when loading models from HuggingFace. This allows a remote attacker to achieve arbitrary Python code execution by convincing a user to run `ilab train/download/generate` with a specially crafted malicious model from the HuggingFace Hub. This vulnerability can lead to complete system compromise.
|
| Publication Date |
April 22, 2026, 11:17 p.m. |
| Registration Date |
April 25, 2026, 4:05 a.m. |
| Last Update |
April 23, 2026, 6:23 a.m. |
|
CVSS3.1 : HIGH
|
| スコア |
8.8
|
| Vector |
CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H |
| 攻撃元区分(AV) |
ネットワーク |
| 攻撃条件の複雑さ(AC) |
低 |
| 攻撃に必要な特権レベル(PR) |
不要 |
| 利用者の関与(UI) |
要 |
| 影響の想定範囲(S) |
変更なし |
| 機密性への影響(C) |
高 |
| 完全性への影響(I) |
高 |
| 可用性への影響(A) |
高 |
Related information, measures and tools
Common Vulnerabilities List