highCVSS 8.8Vulnerability
CVE-2026-6859
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.
Properties
- summary
- InstructLab Includes Functionality from Untrusted Control Sphere
- severity
- high
- epss_score
- 0.00441
- cvss_score
- 8.8
- ghsa_published
- 2026-04-22T15:31:45Z
- source_url
- https://github.com/advisories/GHSA-rxpq-xgqx-fr7p
- ghsa_updated
- 2026-04-29T22:08:39Z
- ghsa_id
- GHSA-rxpq-xgqx-fr7p
- cve_id
- CVE-2026-6859
- cvss_vector
- CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H
- is_ghsa_only
- false
- epss_percentile
- 0.36879
Related Entities (5)
ENRICHED_BY (1)
→[Source]FIRST EPSS
REPORTED_BY (1)
→[Source]GitHub Advisory Database
VULNERABLE_TO (1)
←[Software]pip/instructlab
AFFECTS (1)
→[Software]pip/instructlab
HAS_WEAKNESS (1)
→[Weakness]Inclusion of Functionality from Untrusted Control Sphere
Explore deeper with Ninja Signal's threat intelligence graph