UEFI network boot flaw CVE-2026-97631: what server owners should do

By the NexusSec team · 4 min read · Published
Short version: CVE-2026-97631 is a critical flaw, rated 9.5, in the network-boot (PXE) code of EDK II, the open-source UEFI firmware many server makers build on. An attacker on the same network can send a crafted message to a machine while it boots over the network, crashing it or, with much more difficulty, running code before Secure Boot checks anything. There is no known exploitation, and the fix depends on your server vendor's firmware. Disable PXE where you do not need it and keep provisioning networks isolated.

Firmware flaws are easy to ignore because they are hard to patch. This one is worth a few minutes, because the main mitigation is simple and the affected feature is often switched on by default. This post reflects what was published up to 1 November 2026.

What happened

On 22 October 2026, TianoCore, the project behind EDK II, published an advisory for CVE-2026-97631, rated 9.5 out of 10. EDK II is open-source UEFI firmware code that many firmware makers use as a base. The flaw is a buffer overflow in the PXE network-boot driver: when it copies ICMP error messages into a fixed-size buffer, the size checks are wrong in both the IPv4 and IPv6 code. According to The Hacker News, the flawed code dates back to 2010.

Who is affected

How serious is it?

The attacker needs no password, but must be on an adjacent network and send a crafted ICMP error message while the target is booting over the network. The straightforward result is a crash that hangs the boot. Running code is much harder and depends on the exact firmware build, but it would happen before Secure Boot and other boot-time checks. As of the reports, there was no known exploitation, and the flaw was not in CISA's Known Exploited Vulnerabilities catalogue.

Fix status

A fix was merged into EDK II's main branch on 15 October, and Microsoft's Project Mu published builds containing it on 26 October, but no stable EDK II release included it as of 28 October. For physical servers, the fix arrives as a BIOS or UEFI update from the hardware vendor, so the timing depends on them.

What to do

  1. List the servers and appliances that use PXE boot, and disable PXE in the firmware settings wherever you do not need it.
  2. Keep provisioning and PXE networks in their own segment, reachable only by the machines that need them. See network segmentation.
  3. Watch your server vendors' firmware advisories and apply BIOS or UEFI updates when they ship.
  4. Update virtual machine firmware packages on your hypervisors when your Linux distribution releases the fix.
  5. Monitor boot-time network traffic on provisioning networks, as the advisory suggests.

The flaw was credited to researchers Alex Matrosov of Anthropic and Dan Holmes-Mitra of Google. Coordinated disclosure had been planned for January 2027, but the details became public early.

Sources

Frequently asked questions

Are our office PCs affected by CVE-2026-97631?

Only machines that boot over the network are exposed. Most office PCs boot from a local disk, so they are not affected unless they are set up for network boot.

Is CVE-2026-97631 being exploited?

As of the reports up to 1 November 2026, there was no known exploitation and it was not in CISA's Known Exploited Vulnerabilities catalogue.

How do we fix it?

Install the BIOS or UEFI update from your server vendor when it ships. Until then, disable PXE boot where you do not need it and keep provisioning networks isolated.

Does Secure Boot protect against it?

No. Code run through this flaw would execute before Secure Boot and other boot-time checks.

Running servers or a small data centre?

Book a call and we will look at how your servers boot and how well your provisioning network is separated from everything else.

Book a free consultation