CVE-2026-21387

Awaiting Analysis Awaiting Analysis - Queue

Protection Mechanism Failure in Intel LLM Library for PyTorch

Publication date: 2026-08-11

Last updated on: 2026-08-12

Assigner: Intel Corporation

Description

Protection mechanism failure for some Intel(R) LLM Library for PyTorch within Ring 3: User Applications may allow an escalation of privilege. Unprivileged software adversary with a privileged user combined with a low complexity attack may enable escalation of privilege. This result may potentially occur via local access when attack requirements are present without special internal knowledge and requires passive user interaction. The potential vulnerability may impact the confidentiality (high), integrity (high) and availability (high) of the vulnerable system, resulting in subsequent system confidentiality (none), integrity (none) and availability (none) impacts.

CVSS Scores

EPSS Scores

Probability:
Percentile:

Meta Information

Published

2026-08-11

Last Modified

2026-08-12

Generated

2026-09-01

AI Q&A

2026-08-12

EPSS Evaluated

2026-08-30

NVD

CVE-2026-21387

EUVD

EUVD-2026-56514

Affected Vendors & Products

Vendor	Product	Version / Range
intel	llm_library_for_pytorch	*

Helpful Resources

Exploitability

CWE

KEV

CWE ID	Description
CWE-693	The product does not use or incorrectly uses a protection mechanism that provides sufficient defense against directed attacks against the product.

Attack-Flow Graph

Executive Summary

This is a privilege escalation vulnerability in Intel's LLM Library for PyTorch. It allows unprivileged software to gain higher privileges when combined with a privileged user. The attack requires local access, low complexity, and passive user interaction but no special internal knowledge.

Detection Guidance

This vulnerability is specific to Intel(R) LLM Library for PyTorch and requires local access with privileged user interaction. Detection may involve checking installed versions of the library and monitoring for unusual privilege escalation patterns. No specific commands are provided in the context.

Impact Analysis

An attacker could exploit this to gain elevated privileges on your system, potentially leading to unauthorized access, data manipulation, or system disruption. The impact includes high confidentiality, integrity, and availability risks to the vulnerable system.

Compliance Impact

This vulnerability could lead to unauthorized access or data breaches, violating GDPR's confidentiality requirements and HIPAA's integrity and availability standards. Compliance may be compromised due to potential loss of data protection controls.

Mitigation Strategies

Apply Intel's official patches or updates for the Intel LLM Library for PyTorch. Review and restrict user privileges to minimize attack surface. Monitor system logs for suspicious activity related to privilege escalation attempts.

Hi! I’m here to help you understand CVE-2026-21387. Ask me anything about the vulnerability, its impact, or mitigation strategies.

0/70