ISMS

Secure Boot Certificates Are Expiring: What Admins Need to Check Now

TL;DR
  • Three Microsoft certificates from 2011 expire in 2026: KEK CA 2011 on June 24, UEFI CA 2011 on June 27, and Windows Production PCA 2011 on October 19. As of today, the last deadline is just under two weeks away.
  • Devices without the 2023 certificates keep booting and keep installing Windows updates. They stop receiving new protections for the early boot process, meaning no updates to the boot manager, no revocation lists and no fixes for newly discovered boot-level vulnerabilities.
  • Every device with Secure Boot enabled is affected: Windows 11, Windows 10 with ESU, Windows Server from 2012, virtual machines and Linux in dual boot. Devices with Secure Boot disabled do not receive the new certificates.
  • You check the status with Windows Security, the registry value UEFICA2023Status and events 1801 and 1808. You roll out via Windows Update, Intune, Group Policy or the registry value AvailableUpdates set to 0x5944.
  • Even after October 19 you can still install the new certificates as long as the device boots and installs updates. Microsoft recommends updating at least four test devices per hardware model before a broad rollout.

On October 19, 2026, the certificate Microsoft has used since 2011 to sign the Windows boot manager expires. Two other Secure Boot certificates from 2011 already expired in June, and you only know whether your devices have the new successors if you have actually looked. This article is a work plan for admins and IT managers in mid-sized companies: What really happens, which devices do you need to touch, how do you roll out, and what do you file in the ISMS?

What expires when, and what really happens?

Secure Boot is the UEFI feature that only runs software with a valid signature at startup. The firmware checks those signatures against certificates stored inside it: The Key Exchange Key (KEK) authorizes changes to the databases, the signature database (DB) holds the trusted certificates, and the revocation list (DBX) holds forbidden ones. Since Windows 8, practically all Windows devices carry the same set of Microsoft certificates from 2011 in KEK and DB.

Microsoft is renewing this set. The dates come from the Microsoft Support page on the Secure Boot certificate expiration (page status: May 2026):

2011 certificate Expires on Successor Stored in Purpose
Microsoft Corporation KEK CA 2011 06/24/2026 Microsoft Corporation KEK 2K CA 2023 KEK Signs updates to DB and DBX
Microsoft UEFI CA 2011 06/27/2026 Microsoft UEFI CA 2023 and Microsoft Option ROM UEFI CA 2023 DB Third-party boot loaders and option ROMs
Microsoft Windows Production PCA 2011 10/19/2026 Windows UEFI CA 2023 DB Signs the Windows boot manager

The UEFI CA 2011 is split into two certificates during renewal. One signs third-party boot loaders, the other their option ROMs, meaning firmware drivers for graphics or network cards, for example. October 19 happens to fall on a Monday. If you plan restarts, the weekend before is part of your window.

What happens after October 19?

Less than some warnings suggest, but more than you should ignore. Microsoft describes the consequences like this: Devices that have not received the 2023 certificates keep starting and running normally, and the usual Windows updates keep installing. They can no longer receive new security mechanisms for the early boot process, though. That includes updates to the Windows boot manager and the Secure Boot databases, revocation lists, and fixes for newly discovered vulnerabilities at boot level. Microsoft names fixes for BitLocker bypasses as one example.

The reason: An expired certificate may no longer sign anything new. Protection therefore ages gradually, a little more with every newly discovered boot vulnerability. That is exactly the kind of risk that slips through vulnerability management: no outage, no error message, just a growing gap.

Two more things you should know:

  • October 19 is not a hard cutoff for the technology. According to the Microsoft FAQ, you can still install the new certificates after expiry as long as the device boots and can install updates. The date is still a good anchor, because you want pilot, rollout and evidence done by then.
  • Turning off Secure Boot is not a solution. Microsoft explicitly advises against it. A device without Secure Boot does not get the new certificates and has no protection against bootkits.

Who is affected? Windows, servers, VMs and Linux in dual boot

As a running example, take a company of about 110 employees: 95 notebooks and desktops on Windows 11, 12 devices on Windows 10 with ESU, six Windows servers on Hyper-V and two developer notebooks with Ubuntu in dual boot. IT manages the clients with Intune and the servers via Group Policy. That is a typical picture in mid-sized companies.

Windows 11 and Windows 10

Every device with Secure Boot enabled is affected, which includes most devices bought since 2012. For Windows 11, Microsoft delivers the certificates through Windows Update. There are two aids Microsoft calls "assists". The first delivers the new certificates through the monthly updates to device groups where Microsoft has observed enough successful updates (high-confidence buckets). It is on by default. The second, the Controlled Feature Rollout, handles the change for devices that are enrolled and send diagnostic data. According to the Microsoft FAQ, both are an aid, not a guarantee: The customer remains responsible for the certificates.

Windows 10 only gets the certificates while it is supported. Support ended on October 14, 2025, and since then updates are only available with Extended Security Updates (ESU) or on an LTSC version. According to Microsoft's lifecycle FAQ, the first ESU year for Windows 10 ends on October 13, 2026. If you keep running Windows 10, you need the second-year extension, otherwise those devices stop getting not only boot protection but all updates. In our example these are the 12 older devices: Check the ESU status first, before you think about certificates.

Windows Server and virtual machines

Microsoft's documentation lists Windows Server 2012 and 2012 R2 (both with ESU), 2016, 2019, 2022 and 2025. Servers have two particularities. First, Windows Security on servers does not show the new status indicators by default, so you need central monitoring. Second, many servers run virtualized.

For virtual machines it depends on the virtual firmware. Microsoft names two paths: The maker of the virtualization (Hyper-V, VMware, Azure, AWS and others) ships new virtual firmware with the 2023 certificates, which applies to newly created VMs. For long-lived VMs, Windows installs the certificates just like on a physical device, provided the virtual firmware supports Secure Boot updates. Two known issues from the Microsoft documentation matter for Hyper-V:

  • On some Hyper-V VMs the KEK update fails with event 1795 and the message "The media is write protected". The fix is in the updates from March 10, 2026, and for Windows Server 2025 from April 14, 2026. You have to install it on host and guest. From inside the guest the event tells you where it is missing: According to Microsoft, 1795 reports that the host still needs the update, 1803 that the guest needs it.
  • On Azure Trusted Launch (Gen2) the KEK update can also hang with event 1795. According to Microsoft, as of July 2026 no customer action is required there, and a resolution follows through future updates.

Linux in dual boot

Linux distributions also boot through a Microsoft-signed shim, and until now it carried the signature of the UEFI CA 2011. Once the Microsoft UEFI CA 2023 is in the DB, the firmware can also verify new shims signed with the 2023 certificate. If a device stays on the old DB, the system keeps booting but no longer receives new shim updates and revocations. The distributions maintain their own guidance, so check the documentation of Ubuntu, Fedora or your distribution.

For verification, Microsoft gives these commands for Linux VMs, which you can also use on dual-boot notebooks. Each command must return a matching line:

mokutil --db | grep "UEFI CA 2023"
mokutil --kek | grep "KEK 2K CA 2023"

The order matters: First the certificates go into the firmware, after that the shim updates from the distribution are safe, according to Microsoft. In our example the two developer notebooks therefore go into the first pilot group.

Devices with Secure Boot disabled

Where Secure Boot is switched off, the firmware does not receive the new certificates, and the devices stay vulnerable to bootkits. The expiry changes nothing about that, since the protection did not exist before either. A disabled Secure Boot is still a finding and belongs in your inventory.

Checking the status: Which devices still have the old certificates?

Before you roll out, you need a picture of your fleet. The inventory is the foundation: manufacturer, model, BIOS version and date. If you have maintained your IT asset management, you already have this data. If not, this is the occasion to collect it.

Checking a single device

On one device there are three ways, from simplest to most precise.

Windows Security. Under Start, Settings, Privacy & security, Windows Security, Device security, the Secure boot section shows the status. Since April 2026 Windows also shows the certificate status there. A green checkmark alone is not enough, look for the text "Secure Boot is on and all required certificate updates have been applied." Yellow stands for a recommendation, for example because of a firmware limit, red for a boot security update this device cannot receive. On managed Windows clients and on servers these indicators are off by default. You switch them on with the registry value HideSecureBootStates set to 0, under HKLM\SOFTWARE\Policies\Microsoft\Windows Defender Security Center\Device security. For the fleet you need a central evaluation anyway.

PowerShell. In a PowerShell with administrator rights, you first check whether Secure Boot is running at all:

Confirm-SecureBootUEFI

The command returns True if Secure Boot is on. Then you read the DB in readable form. The -Decoded parameter was added with the monthly updates of April 14, 2026:

Get-SecureBootUEFI -Name db -Decoded

In the output you look for the subject CN=Windows UEFI CA 2023. If only CN=Microsoft Windows Production PCA 2011 is in the list, the firmware is still on the old state. Microsoft's sample output also shows the validity: The PCA 2011 is valid until 2026-10-19, the Windows UEFI CA 2023 until 2035-06-13. The same goes for the KEK: With -Name KEK instead of -Name db you look for Microsoft Corporation KEK 2K CA 2023.

Registry and Event Viewer. Windows records progress under HKLM\SYSTEM\CurrentControlSet\Control\SecureBoot\Servicing. The value UEFICA2023Status is NotStarted, InProgress or Updated. On an error, UEFICA2023Error holds a code other than 0. Two events matter in the System event log: 1801 is an error event and means the new certificates have not been applied yet. 1808 is an informational event and confirms that the firmware has the new certificates. For the status query you use UEFICA2023Status, not WindowsUEFICA2023Capable.

Evaluating the whole fleet

For more than a handful of devices you need a script that collects these values. Microsoft provides a sample inventory script. On devices with the updates from May 12, 2026, it sits in the folder %systemroot%\SecureBoot\ExampleRolloutScripts. It reads registry values, hardware data and events 1801 and 1808 and outputs the result as JSON. In Intune this script runs as the detection script of a remediation, without changing anything on the devices. You see the results in the Intune portal and export them as CSV.

For planning, the confidence level that Microsoft assigns to each device matters:

Confidence Meaning What you do
High Confidence Microsoft has validated this device class as safe Roll out the update
Under Observation Microsoft is still collecting data Wait for classification or test yourself
No Data Observed Device class is unknown to Microsoft Test yourself and plan the rollout
Temporarily Paused Known compatibility issue Check for an OEM BIOS update, wait for Microsoft
Not Supported Hardware or platform limitation Document as an exception

The result is a table with one row per device and the columns model, BIOS, confidence, status. You build the pilot groups from it.

How do you roll out the 2023 certificates?

Microsoft describes several paths, and the documentation advises against mixing them on the same device. Which one fits depends on how you manage your devices. The certificates also arrive with the cumulative updates, so you need a working patch management process.

Path 1: Windows Update and the Microsoft assists

The most convenient path is to do nothing special and distribute the monthly updates cleanly. Devices in high-confidence buckets then get the certificates automatically. The registry value HighConfidenceOptOut controls this: 0 or absent means participation, 1 means opt-out. The Controlled Feature Rollout is activated through MicrosoftUpdateManagedOptIn (0 or absent means off, any non-zero value means on) and needs the required diagnostic data.

If you deliberately delay updates through rings, the certificates arrive later accordingly. For devices that are not in a high-confidence group or do not send diagnostic data, do not rely on the assists.

Path 2: Intune

In Intune you create a new policy under Devices, Manage devices, Configuration. The platform is Windows 10 and later, the profile type is the settings catalog. In the settings picker you search for "Secure Boot" and find three settings that correspond to the registry values:

  • Enable Secureboot Certificate Updates starts the rollout on the device (corresponds to AvailableUpdates).
  • Configure High Confidence Opt-Out controls whether the automatic updates run through Windows Update.
  • Configure Microsoft Update Managed Opt In enrolls devices in the Controlled Feature Rollout.

Assign the policy to a device group, not to all devices at once. A known issue affected Pro editions: Until the end of January 2026, Secure Boot settings were rejected there with error code 65000. Microsoft fixed this through the licensing service, and older licenses renew themselves monthly. If you still see the error on Windows 11 23H2 Pro, the update of April 14, 2026 (KB5082052) helps.

Path 3: Group Policy

For domain PCs and servers there is a Group Policy. First you download the administrative templates (ADMX) that Microsoft provides. Then you find the settings under Computer Configuration, Administrative Templates, Windows Components, Secure Boot:

  • Enable Secure Boot Certificate Deployment starts the rollout (registry: AvailableUpdates).
  • Automatic Certificate Deployment via Updates controls automatic distribution through Windows Update. Mind the logic: "Enabled" blocks automatic distribution here.
  • Certificate Deployment via Controlled Feature Rollout enrolls devices in the rollout run by Microsoft.

Two notes from the Microsoft documentation matter in practice. First: If you remove the policy again, the registry value stays. Second: Once certificates are written to the firmware, you cannot remove them from Windows, only through the firmware interface. The policy is therefore not a switch you simply flip back when problems occur.

Path 4: Registry and script

If you use neither Intune nor Group Policy, the registry path remains. On every target device you set the value AvailableUpdates to 0x5944. That is the value that triggers all necessary steps: the new certificates into the DB, the new KEK and the new boot manager. To test on a single device, you run these commands from Microsoft's documentation one by one in a PowerShell with administrator rights:

reg add HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Secureboot /v AvailableUpdates /t REG_DWORD /d 0x5944 /f
Start-ScheduledTask -TaskName "\Microsoft\Windows\PI\Secure-Boot-Update"

The value first drops to 0x4100. Then you restart the device and run the scheduled task once more so Windows installs the new boot manager. Normally the task runs at system startup and every 12 hours after that. The trigger itself does not force a restart, but the device has to restart at some point. For planning, Microsoft gives about 48 hours and at least one restart per device.

Windows works through the steps in a fixed order: first the Windows UEFI CA 2023 into the DB, then (if the UEFI CA 2011 is in the DB) the Microsoft UEFI CA 2023 and the Option ROM UEFI CA 2023, then the KEK, and finally the boot manager signed with the Windows UEFI CA 2023. If a step fails, its bit stays set and the task retries on the next run.

Firmware from the manufacturer

The KEK step needs the device manufacturer's cooperation. Only whoever controls the Platform Key (PK) may change the KEK, and the PK belongs to the OEM. The manufacturer has to give Microsoft a KEK signed with the PK, which Windows then ships with its updates. If it is missing, the update hangs at this point: AvailableUpdates keeps the KEK bit (0x0004), and event 1803 appears repeatedly in the System event log. Microsoft states there is no supported manual path for this condition. It mostly affects older devices for which the manufacturer no longer maintains firmware. Event 1795, in contrast, points to a firmware error while writing the variable. Then you check whether a BIOS update exists.

Newer devices built in the last one to two years may already have the 2023 certificates from the factory but not yet the new boot manager installed. According to Microsoft, this last step is also critical, so check devices that look "new enough" too.

Where the firmware does not process the updates correctly and no BIOS update exists any more, Microsoft recommends replacing the device. Until then it is a documented exception, more on that below.

How do you avoid BitLocker recovery surprises?

BitLocker seals its key against measurements the TPM records at startup. If the Secure Boot configuration or the boot manager changes, a device may ask for the recovery key at the next start. In its troubleshooting guide, Microsoft describes three scenarios you should know before you roll out.

One-time recovery after the update. On the first start after the update, the firmware does not yet report the new values when Windows reseals BitLocker. The device therefore asks for the recovery key once. On the next start the values match and nothing happens any more. Still check whether a newer BIOS exists for the model.

Repeated recovery with PXE as the first boot device. If the network boot device is set ahead of the disk, the PXE start fails and the firmware falls back to the boot manager on the disk. The TPM then measures two different signature chains in one startup: the PXE image with the Microsoft UEFI CA 2011 and the local boot manager with the Windows UEFI CA 2023. BitLocker finds no stable measurements and asks for the key at every start. Remedy: change the boot order, disable PXE, or move the PXE infrastructure to a 2023-signed Windows boot loader. If you reinstall devices over PXE, this belongs in your pilot phase.

Boot failure after resetting to firmware defaults. If someone resets Secure Boot to factory defaults in the BIOS, that clears the databases. If the device already runs the 2023-signed boot manager, the firmware lacks the certificate for it and the device no longer boots. Microsoft describes a rescue via USB stick: From a second PC with the July 2024 update or newer, you copy SecureBootRecovery.efi from the Windows boot directory to a FAT32 stick as \EFI\BOOT\bootx64.efi and boot the problem device from it. The program only writes the Windows UEFI CA 2023 back into the DB. Afterwards you need a current BIOS and must install the remaining certificates again. The simple rule: Do not load BIOS defaults unless the manufacturer ships defaults that include the 2023 certificates.

That leads to the preparation. Before the pilot, check that the recovery keys of all devices are stored in Entra ID or Active Directory and that the help desk can access them. Tell users that a prompt may appear once after the restart and whom to call. And do not let home-office devices restart on a Friday evening. A solid encryption strategy settles the key question up front, because without a findable recovery key a restart turns into a data loss scenario.

Pilot groups: How to proceed in the remaining days

Microsoft recommends working in stages: a small group first, then more, and at least four test devices per category (manufacturer, model, firmware version) before you roll out broadly. You are changing systems in the boot path, so this belongs in your change management process, with approval, rollback plan and documentation.

For our example company the plan looks like this:

Period Wave Devices Goal
Oct 6 to 7 Inventory All devices Record status, models, BIOS level, confidence; check recovery keys
Oct 8 to 9 Pilot IT devices, 2 Linux dual-boot notebooks, at least 4 devices per model, 1 Hyper-V server Test the rollout; watch BitLocker, PXE, Linux boot
Oct 12 to 14 Wave 1 Standard models in high confidence, about 60 clients Rollout via Intune; monitor 1801/1808
Oct 14 to 16 Wave 2 Remaining clients, servers after maintenance window Rollout; clear failure cases one by one
Oct 19 Deadline All Evaluate status, document exceptions with date and justification

During the waves you watch three things: how many devices have UEFICA2023Status at Updated, how many report event 1808, and which devices show UEFICA2023Error with a non-zero value. Failures often come down to firmware, and then the manufacturer's site helps, not another attempt. Servers need maintenance windows, and for VMs the host updates are part of the job.

Evidence in the ISMS: What you document

The certificate change is a small example of vulnerability and change management, with a clear deadline and a clear result. Auditors want to see the question "Where do we stand?" backed by evidence, not told as a story. In ISO 27001 this maps to control A.8.8 (management of technical vulnerabilities) and A.8.32 (change management).

A solid body of evidence has five building blocks:

  1. Risk. You create a risk: "Devices without the 2023 Secure Boot certificates no longer receive boot security updates." You rate likelihood and impact with your usual matrix. You link the risk to the affected assets.
  2. Measure. The rollout is tracked as a measure with an owner, a due date and an effectiveness check. It is effective when the evaluation shows UEFICA2023Status at Updated on all supported devices.
  3. Inventory with status. The device list with model, BIOS, confidence and status on the deadline is your main evidence. Export it on October 19 and monthly afterwards.
  4. Exception list. For every device that does not get the certificates you need a justification, a compensating measure (for example replacement by quarter X, no access to sensitive systems) and a risk acceptance by the responsible person.
  5. Change records. Approval, pilot log, rollback plan and closing report for the change.

ISMS Lite brings these building blocks together: The risk is linked to the assets, the measure carries an owner and a due date, and exceptions can be maintained as separate risks with a review date. If you already hold device data in an inventory system, you can maintain it through the API instead of transferring it by hand.

The change is also a good occasion to review your inventory of cryptographic items. Certificates have an expiry date, and in many companies nobody knows all of those embedded in firmware, devices and services. How to build such a certificate and crypto inventory is covered in a separate article, and the rules for algorithms and lifecycles belong in your cryptography policy.

Checklist: Secure Boot certificates by October 19

Inventory

  • All devices inventoried with manufacturer, model, BIOS version and date
  • Secure Boot status collected per device (Confirm-SecureBootUEFI)
  • Devices with Secure Boot disabled or legacy BIOS documented as a finding
  • UEFICA2023Status and UEFICA2023Error evaluated centrally (Intune remediation, script or management tool)
  • Confidence level recorded per device model
  • Windows 10 devices: ESU status checked, second-year extension clarified
  • Servers and VMs inventoried, Hyper-V hosts and guests with updates from March and April 2026 respectively

Preparation

  • BitLocker recovery keys of all devices verified in Entra ID or Active Directory
  • PXE boot order known on all devices, PXE infrastructure checked
  • Current manufacturer BIOS versions checked per model
  • Administrative templates for Group Policy loaded or Intune policy prepared
  • Change request with rollback plan approved
  • Users and help desk informed about the possible one-time BitLocker prompt

Pilot and rollout

  • Pilot with at least four devices per model completed
  • Linux dual-boot devices tested (mokutil --db, mokutil --kek)
  • Rollout in waves, restarts scheduled
  • Events 1801 and 1808 and UEFICA2023Error monitored
  • Failure cases clarified one by one, manufacturer support involved
  • BIOS defaults not loaded unless the manufacturer ships new defaults

Evidence

  • Risk and measure created in the ISMS and linked to assets
  • Device list with status exported and filed on the deadline
  • Exceptions documented with justification, compensating measure and risk acceptance
  • Effectiveness check scheduled (monthly reconciliation)

Common mistakes in the certificate change

Relying on Windows Update alone. The assists are aids, not a guarantee, and Microsoft says so itself. Devices without diagnostic data, with delayed updates or in unknown device groups do not reach the new state on their own. If you do not measure, you only notice when the next boot fix fails to arrive on those devices.

Rolling out to all devices at once. Firmware is the risk, not Windows. A model with faulty DB handling can fail to boot after the restart, and with a hundred devices at once you have a hundred cases at once. Two pilot devices per model are not enough, Microsoft recommends four.

Resetting the BIOS to factory settings. This is the classic help desk reflex for boot problems. If the device already runs the 2023 boot manager, the reset deletes the matching certificate and the device no longer starts. Put that in the help desk's work instructions.

Turning off Secure Boot to get some peace. That removes the warning and removes the protection. Microsoft explicitly advises against it. If at all, it is a time-limited, documented exception for a single device and not a tool for the fleet.

Exceptions without documentation. A device that does not get the certificates is not an error but a decision. Without a justification, a date for replacement and a risk acceptance, it is an open vulnerability without an owner in the audit.

Conclusion: The deadline is soft, the evidence is not

The expiry does not take any device down, but it splits your fleet into devices that can accept future boot security updates and devices that no longer can. Start this week with the inventory and the pilot, because what runs short is the time for restarts and firmware cases, not the effort for the setting itself. If you track the change in the ISMS as a risk with a measure, a device list and exceptions, you end up after the deadline with not only secure devices but also the proof.

Further reading

Document the certificate change as a measure

In ISMS Lite you create the Secure Boot change as a measure with an owner and a due date, link it to the risk and the affected assets, and record the effectiveness check as evidence.

Install now