EUVD-2026-24736
GHSA-pqmg-c2j8-fq92
Severity by source
AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:N
Primary rating from NVD.
CVSS VectorNVD
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:N
Lifecycle Timeline
5DescriptionCVE.org
A flaw was found in InstructLab. A local attacker could exploit a path traversal vulnerability in the chat session handler by manipulating the logs_dir parameter. This allows the attacker to create new directories and write files to arbitrary locations on the system, potentially leading to unauthorized data modification or disclosure.
AnalysisAI
Path traversal in InstructLab's chat session handler enables local authenticated attackers to write files to arbitrary filesystem locations by manipulating the logs_dir parameter. Red Hat Enterprise Linux AI 3 deployments are confirmed affected. CVSS 7.1 (High) reflects significant confidentiality and integrity impact, though exploitation requires local access and low-level privileges. No active exploitation (CISA KEV) or public proof-of-concept identified at time of analysis. EPSS data not available, suggesting limited immediate widespread exploitation risk despite high severity rating.
Technical ContextAI
InstructLab is Red Hat's AI model training and tuning framework for large language models, part of the RHEL AI 3 ecosystem. The vulnerability resides in the chat session log handling component, exploiting CWE-22 (Path Traversal) weakness. Attackers can manipulate the logs_dir parameter without proper sanitization or validation, allowing directory traversal sequences (../, absolute paths) to escape intended logging directories. The flaw affects the session persistence layer where chat interactions are recorded, enabling write operations outside the application's designated data directories. CPE strings confirm impact across multiple RHEL AI 3 components, suggesting the vulnerable code is integrated into the platform's chat interface or CLI tooling.
RemediationAI
Apply vendor-released security updates for Red Hat Enterprise Linux AI 3 as published in Red Hat Security Advisories. Consult https://access.redhat.com/security/cve/CVE-2026-6855 for patched package versions and installation instructions via dnf/yum package managers. Red Hat Bugzilla 2460013 (https://bugzilla.redhat.com/show_bug.cgi?id=2460013) may contain additional technical details and interim workarounds. Until patches are deployed, implement compensating controls: restrict filesystem permissions on directories accessible to InstructLab processes using strict ACLs (trade-off: may break legitimate logging functionality), enforce mandatory access controls via SELinux targeted policies to confine InstructLab write operations to designated paths (requires custom policy development), and monitor file creation events outside expected log directories using auditd rules (detection-only, does not prevent exploitation). Review and rotate credentials for accounts with InstructLab access, as exploitation requires authenticated local access.
Remote code execution in InstructLab affects Red Hat Enterprise Linux AI 3 when users download or train models from Hugg
A remote attacker can trigger a heap out-of-bounds write in FFmpeg's DVD subtitle parser by providing a crafted MPEG-PS/
Use-after-free in FFmpeg's RASC video decoder exposes Red Hat Enterprise Linux AI 3 and Red Hat OpenShift AI deployments
Image input manipulation in vLLM's multimodal preprocessing pipeline allows remote, unauthenticated network attackers to
Same weakness CWE-22 – Path Traversal
View allSame technique Path Traversal
View allVendor StatusVendor
Share
External POC / Exploit Code
Leaving vuln.today