News › security

AMD Memory Controllers Exploit Bypasses Protection, Poses New Security Risks

By Zayden R., August 14, 2026

Christopher Domas reveals a method to bypass memory protection using AMD's memory controllers. This could allow kernel-level code to manipulate CPU instructions, posing potential security risks.

Christopher Domas has published an eye-opening proof of concept detailing how AMD's memory controllers can be manipulated to bypass memory protection. By exploiting the bank swizzle mode, Domas demonstrates the ability to read or write arbitrary data, including sensitive components like CPU microcode definitions and memory belonging to the platform security processor.

This revelation is significant for engineers, as it highlights a potential vulnerability in AMD's architecture that could be exploited for malicious purposes. While the behavior is documented in AMD's manual, the ability to access and alter supposedly immutable parts of the computer's firmware without causing a system crash seems to be an unintended byproduct.

The exploit requires kernel-level privileges to enable bank swizzle mode, meaning the immediate risk to general software users is minimal. However, the method could eventually find its way into the toolkit of those with nefarious intentions, emphasizing the importance of vigilance in system security.

For engineers and system administrators, this discovery underscores the need for robust security measures and the continuous monitoring of systems for unauthorized access attempts. While this specific exploit may not yet be widely used, it's a reminder of the ever-present need to stay ahead in the security game.

Sources

Practice this on a real machine

The Linux Camp teaches these topics as hands-on labs on real virtual machines, verified as you type.