---
id: "CVE-2026-0004"
page: "https://infrastack.ai/live/CVE-2026-0004"
kind: "vulnerability"
aliases: ["CVE-2026-0004"]
title: "AMD processors: transient execution flaw leaks L1 cache data across privilege boundaries"
severity: "medium"
cvss: 5.7
epss: null
exploited: false
status: "open"
category: "infrastructure"
products: [{"name": "AMD processors", "fixed": [], "affected": "", "ecosystem": "vendor"}]
ecosystems: ["vendor"]
tags: ["amd", "cpu", "infrastructure", "multi-tenant", "virtualization"]
not_affected_if: ["no-untrusted-code-execution", "non-amd-hardware"]
published_at: "2026-11-12T19:16:45+00:00"
first_seen_at: "2026-11-12T19:40:13+00:00"
updated_at: "2026-11-12T19:41:00+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-0004", "first_seen_at": "2026-11-12T19:40:13+00:00"}]
exploit_sources: []
---

# AMD processors: transient execution flaw leaks L1 cache data across privilege boundaries

Link for people: https://infrastack.ai/live/CVE-2026-0004 (this .md is for agents)

Some AMD processors have a transient execution flaw that lets code running on the system infer data held in the L1 data cache. An attacker who can already run code on the machine can read sensitive information across privileged boundaries.

_Summary, check and fix written by a model from the sources; labelled as such._

## Affected

- vendor:AMD processors affected range not listed fixed no fixed version listed

## Timeline

- 2026-11-12T19:40:13+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-0004

## Documents

- [mentions] Google Cloud Security Bulletins: GCP-2026-081: Published: 2026-11-10 Description Description Severity Notes AMD has disclosed three vulnerabilities that affect 5th generation AMD EPYC pro https://docs.cloud.google.com/support/bulletins/index#gcp-2026-081

## From the sources

A transient execution vulnerability in some AMD processors may allow an attacker with the ability to execute code on the system to infer data in the L1D cache, potentially resulting in the leakage of sensitive information across privileged boundaries.

---

_Free to read, with fair use of about 60 pages a minute. For higher limits or API access: sales@infrastack.ai_