AMD allegedly has its own Spectre-like security flaws

Researchers say they've found 13 flaws in AMD's Ryzen and EPYC chips, which could let attackers install malware on highly guarded parts of the processor.

Alfred Ng Senior Reporter / CNET News
Alfred Ng was a senior reporter for CNET News. He was raised in Brooklyn and previously worked on the New York Daily News's social media and breaking news teams.
Alfred Ng
6 min read

Researchers say they've discovered critical security flaws in AMD chips that could allow attackers to access sensitive data from highly guarded processors across millions of devices.

Particularly worrisome is the fact that the alleged vulnerabilities lie in what's designed to be the secure part of the processors -- typically where your device stores sensitive data like passwords and encryption keys. It's also where your processor makes sure nothing malicious is running when you start your computer.

The majority of these reported vulnerabilities would require administrative access to work, meaning an attacker would first need to have control of your machine through some type of malware. But even with the need for administrative access, putting the malware on the secure processor itself creates a higher potential for damage than a normal attack would. 

CTS-Labs, a security company based in Israel, announced Tuesday that its researchers had found 13 critical security vulnerabilities that would let attackers access data stored on AMD's Ryzen and EPYC processors, as well as install malware on them. Ryzen chips power desktop and laptop computers, while EPYC processors are found in servers.

"At AMD, security is a top priority and we are continually working to ensure the safety of our users as new risks arise," an AMD spokesman said. "We are investigating this report, which we just received, to understand the methodology and merit of the findings."

The researchers gave AMD less than 24 hours to look at the vulnerabilities and respond before publishing the report. Standard vulnerability disclosure calls for at least 90 days notice so companies have time to address flaws properly. Disclosing a vulnerability to the public without first giving a company enough time to fix it can be irresponsible, as it leaves the flaw open for attackers to exploit. Google's researchers gave Intel six months to fix issues related to Spectre and Meltdown.

On Wednesday, CTS-Labs Chief Technology Officer Ilia Luk-Zilberman released a statement defending his decision regarding the unusual disclosure method. 

"I think that a better way would be to notify the public on day zero that there are vulnerabilities and what is the impact. To notify the public and the vendor together. And not to disclose the actual technical details ever unless it's already fixed," Luk-Zilberman said. 

Critics have also pointed out that along with the unorthodox disclosure, CTS-Labs' legal disclaimer mentions a potential conflict of interest. 

"Although we have a good faith belief in our analysis and believe it to be objective and unbiased," the disclaimer says, "you are advised that we may have, either directly or indirectly, an economic interest in the performance of the securities of the companies whose products are the subject of our reports."

In response to an email about the disclaimer, CTS-Labs said it doesn't have "any investment (long or short) in Intel or AMD."

The report of the vulnerabilities comes after the emergence of Meltdown and Spectre, security flaws in Intel and Arm chips, which affected a huge number of PCs dating back two decades. According to researcher Statista, 77 percent of computer processors are from Intel, while AMD accounts for 22 percent.

When the Meltdown and Spectre flaws were revealed in January, AMD said that because of design differences, its chips weren't affected.

These new security vulnerabilities break down into four categories, according to CTS-Labs co-founder and Chief Financial Officer Yaron Luk-Zilberman. All essentially allow an attacker to target the secure segment of a processor, which is crucial to protecting the sensitive information on your device.

"You're virtually undetectable when you're sitting in the secure processor," the CFO, who previously ran a hedge fund, NineWells Capital Partners, said of the flaws. "An attacker could sit there for years without ever being detected."

Security researchers also criticized the white paper published by CTS-Labs for lacking any technical details describing the vulnerabilities. CTS-Labs said it sent the technical report to Dan Guido, an independent security researcher and the CEO of Trail of Bits.

Guido said the company sent him the details last week, and added that the threats were legitimate.

Guido also said CTS-Labs paid him the company's "week rate for the work." Reuters reported that CTS-Labs paid Trail of Bits about $16,000 for the review. CTS-Labs was founded in 2017, with no history in cybersecurity and six employees, according to its CFO.

Here's a breakdown of the reported flaws:

Master Key

When a device starts up, it typically goes through a "secure boot" process. It uses your processor to check that nothing on your computer has been tampered with, and launches only trusted programs.

The Master Key vulnerability gets around this startup check by installing malware on the computer's BIOS, part of the computer's system that controls how it starts up. After infection, Master Key allows attackers to install malware on the secure processor itself, meaning they'd have complete control over what programs are allowed to run during the startup process.

From there, the vulnerability also allows attackers to disable security features on the processor.


This vulnerability specifically affects AMD's Ryzen chips, CTS-Labs said, and would allow malware to completely take over the secure processor.

That would mean being able to access protected data, including encryption keys and passwords. These are regions on the processor that a normal attacker wouldn't be able to access, according to the researchers.

If attackers could bypass the Windows Defender Credential Guard, they could use the stolen data to spread their assault to other computers within a network. Credential Guard is a feature for Windows 10 Enterprise that stores your sensitive data in a protected section of the operating system that normally can't be accessed.

"The Windows Credential Guard is very effective at protecting passwords on a machine and not allowing them to spread around," Luk-Zilberman said. "The attack makes spreading through the network much easier."


Like Ryzenfall, Fallout also lets attackers access protected data sections, including Credential Guard, CTS-Labs said. But this vulnerability affects only devices using AMD's EPYC secure processor. In December, Microsoft announced a partnership in which its Azure Cloud servers use EPYC processors.

"Windows has a customer commitment to investigate reported security issues, and proactively update impacted devices as soon as possible," a Microsoft spokesperson said. "Our standard policy is to provide solutions via our current Update Tuesday schedule."

These chips are used for data centers and cloud servers, connecting computers used by industries around the world. If attackers used the vulnerabilities described in Fallout, they could steal all the credentials stored and spread across the network.

"These network credentials are stored in a segregated virtual machine where they can't be accessed by standard hacking tools," said CTS-Labs CEO Ido Li On. "What happens with Fallout is that this segregation between virtual machines [is] broken."

Segregated virtual machines are portions of your computer's memory split off from the rest of the device. Researchers use them to test malware without infecting the rest of the computer. Think of it as a computer inside your computer.

With Credential Guard, sensitive data is stored there and protected so that if your computer is infected by normal malware, the malware can't access that data.


Chimera comes from two different vulnerabilities, one in firmware and one in hardware.

The Ryzen chipset itself allows malware to run on it, CTS-Labs said. Because Wi-Fi, network and Bluetooth traffic flows through the chipset, an attacker could use that to infect your device, the researchers said. In a proof-of-concept demonstration, they said, it was possible to install a keylogger, which would allow an attacker to see everything typed on an infected computer.

The chipset's firmware issues mean an attack can install malware onto the processor itself.

"What we discovered is what we believe are very basic mistakes in the code," said Uri Farkas, CTS-Labs vice president of research and design.

What should I do?

It's unclear how long it would take to fix these issues. CTS-Labs said it hasn't heard back from AMD. The researchers said it could take "several months to fix." The vulnerabilities in the hardware can't be fixed.

Intel and Microsoft are still managing patches for Meltdown and Spectre, and the fixes have ended up causing problems, including slower performance on affected computers. These new vulnerabilities could mean similar headaches for AMD-powered devices.

"Once you're able to break in to the security processor, that means most of the security features offered are broken," Li On said. 

First published March 13, 7:07 a.m. PT
Updates, 1:22 p.m.:
Adds details from CTS-Labs' legal disclaimer; 2:08 p.m.: Includes CTS-Labs' response to follow-up question about the disclaimer; March 14 at 12:40 p.m.:Adds statement from CTS-Labs' chief technology officer and information on how much Trail of Bits was paid.

Security:  Stay up-to-date on the latest in breaches, hacks, fixes and all those cybersecurity issues that keep you up at night.

Rebooting the Reef: CNET dives deep into how tech can help save Australia's Great Barrier Reef.