Home/Blog/UEFI firmware CVE-2026-97631: EDK II bug runs code before boot
blog

UEFI firmware CVE-2026-97631: EDK II bug runs code before boot

October 29, 2026 · Security News

UEFI firmware CVE-2026-97631 in TianoCore EDK II allows code execution before the OS and Secure Boot

UEFI firmware CVE-2026-97631 is a 9.5-rated memory-corruption bug in TianoCore EDK II, the open-source code most server boot firmware is built from. It lets an attacker run code before the operating system loads and before Secure Boot checks anything. The fix went public ahead of the planned disclosure, so defenders have no quiet window to patch.

Key takeaways

  • CVE-2026-97631, rated 9.5, lets attackers run code in TianoCore EDK II-based UEFI firmware before the OS and Secure Boot.
  • An AMI pull request revealed the bug and its fix ahead of a coordinated disclosure planned for January.
  • Fixes have to come from each OEM or cloud provider. Inventory your firmware and get vendor answers in writing.

What is UEFI firmware CVE-2026-97631?

CVE-2026-97631 is a memory-corruption flaw in an ICMP component of TianoCore EDK II, the reference code hardware makers and cloud providers start from when building UEFI firmware. A UEFI Forum representative last year called it the preeminent implementation of the UEFI specifications.

Ars Technica reports that virtually every independent BIOS vendor, including AMI, Insyde and Phoenix, uses the code extensively, and Microsoft’s hardened fork, Project Mu, also used the code containing the vulnerable component. One upstream bug can surface in many brands at once.

ICMP underpins ping and network error messages, so the bug likely sits in firmware networking code. Nobody has said what an attacker needs to reach it: UEFI network boot, a rogue boot server and a local foothold are all unconfirmed possibilities.

Does CVE-2026-97631 bypass Secure Boot?

In effect, yes. The attacker’s code runs before Secure Boot and the TPM-anchored chain of trust take over, and those controls only check what loads after them. Per Ars, the payload runs once and does not persist, but it can install follow-on payloads that start early on every boot, underneath the OS.

Reinstalling Windows or Linux does not reflash motherboard firmware, and endpoint tools see little of what ran before the OS. A clean antivirus result on a misbehaving server means less than usual. Threat hunting below the OS is the right next step.

Why did the disclosure collapse?

Ars credits the discovery to Alex Matrosov of Anthropic and Dan Holmes-Mitra of Google, who had arranged a coordinated disclosure for January. Then AMI published a pull request revealing the bug details and the code fix. Ars does not say whether this was accidental. Once the discoverers spotted it, TianoCore Project members published a public advisory.

Ars expects advanced attackers to compare the fixed and unfixed code and build working exploits, while the fix crawls onto millions of devices from thousands of hardware makers and cloud providers. It was the second botched disclosure in a week: a critical PHP code-execution flaw went public six days earlier.

What should a business do about the EDK II bug this week?

You cannot apply this patch yourself. Each manufacturer must build the EDK II fix into its own signed firmware, and each cloud provider must roll it out across its hosts.

  • Record firmware versions. On Windows, Get-CimInstance Win32_BIOS; on Linux, dmidecode -t bios. Save the output for comparison after updates.
  • Ask every OEM and cloud provider in writing whether your exact models or instance types are affected and when a fixed build ships. A “not affected” reply should name the product.
  • Ask about network boot. If a server never boots over the network, ask whether disabling UEFI network boot reduces exposure. Until confirmed, it is a stopgap, not the fix.
  • Keep Secure Boot on. It does not stop this bug but still blocks other boot attacks.
  • Check after the reboot that the new firmware version actually loaded.

A vendor bulletin says a fix exists, not that your hardware runs it. To prove that, use firmware security analysis.

FAQ

Will updating Windows or Linux fix CVE-2026-97631?

No. The bug lives in motherboard firmware, so only a firmware update from your hardware maker or cloud provider fixes it.

Is there a patch for CVE-2026-97631?

The upstream fix is public in EDK II, but you need your vendor's signed firmware build that includes it. Ask for the fixed version number and release date.

Should I turn off Secure Boot until a firmware update arrives?

No. Secure Boot does not stop this bug, but it still blocks unsigned bootloaders and other boot attacks.

If a server is behaving in ways you can’t explain and rebuilding the OS hasn’t fixed it, our compromised server recovery team checks below the operating system, starting with the firmware.


← All articles

get in touch

Tell us what you're running

Scoping is free. We reply within one business day, and under 30 minutes for active incidents.