topicsidentity & access3 numbers3 guides

emergency access accounts

The one admin account built to skip every control that can fail. That makes it the login you most need on a bad day and the one attackers would most like on any other.

in one lineAn emergency access account is a top-level admin login kept outside the normal sign-in rules, owned by nobody, so the company can still get in when every other admin path fails.

the numbers

all the numbers →

1what it is

An emergency access account, often called a break-glass account, is the admin login you keep for the day the normal ones stop working. The identity provider you federate with is down. The phone network is out and nobody can answer an MFA prompt. The last person with Global Administrator has left. A new sign-in policy locks every admin out at once.

Microsoft's documentation for Entra ID asks for two or more of these accounts: cloud-only, on the tenant's own onmicrosoft.com address, permanently holding Global Administrator, tied to no individual and to no employee's phone, excluded from Conditional Access policies that block or restrict sign-in, and protected by a passkey on a hardware key kept in a safe. Google's Workspace guidance never uses the words break-glass or emergency access, but asks for the same shape: more than one super admin, each separate from the account used for daily work, protected by security keys, with backup codes printed and stored somewhere safe.

The full setup is in the break-glass guide.

2why it accumulates

An emergency account is built to be ignored. Nobody signs into it on a normal day, so every small change around it goes unchecked.

The credential ages. A common older setup gave the account a long password and an exclusion from the MFA policy, so it could never lock anyone out. If that password sits in a shared vault, everyone who has had access to the vault over the years, former contractors included, may have seen it.

The exclusion group grows. Microsoft recommends one dedicated group for the accounts that skip blocking policies. Whoever can add members to that group can take any account out of your sign-in rules, and nobody looks at the member list.

The recovery details point at a person. Someone registered their own phone for MFA or password reset during setup. When they leave, the account depends on someone who is gone, the situation the founder departure guide describes.

The vendor changes the rules. Microsoft's mandatory MFA page says break-glass accounts must also sign in with MFA once enforcement begins, that exclusions in your own policies no longer apply, and that "There's no way to opt out." Enforcement on the Azure portal, Entra admin center and Intune admin center began in October 2024, and on the Microsoft 365 admin center in February 2025. Google says 2-Step Verification for admins is rolling out over the coming years and admins cannot bypass it. A password-only emergency account now fails on the admin portals it exists for. The vendor login changes guide covers that pattern.

3what it costs you

It fails when you need it. The account was reset, disabled by an inactivity clean-up, blocked by a new policy, or tied to a phone that left with its owner. The outage you planned for becomes a lockout, and the way back runs through the vendor's support process on the vendor's timetable.

It works for someone else. A permanent top-level admin, exempt from blocking policies, with no named owner and no alert, is the account an attacker would design for themselves. Password guessing runs at scale: Microsoft's Digital Defense Report 2024, which is vendor research, says Microsoft blocked 7,000 password attacks per second over the year. The Midnight Blizzard intrusion began with a legacy test account that had no MFA, which is the same pattern of an account kept outside the normal rules.

There is no public data on how many small companies have an emergency account, or how many of those still work. The count that matters is your own: how many accounts hold the top role. Microsoft's best-practices page recommends fewer than five people with Global Administrator, and CISA's secure configuration baseline for Entra ID (MS.AAD.7.1v1) sets a ceiling of eight, emergency accounts included. For accounts that skip MFA without anyone deciding they should, see MFA gaps.

what it looks like when it goes wrong

Three ways this tends to go, each one taken from a real case.

your company
the gap nobody seesThe emergency admin account was set up once, years ago, and nobody has signed into it or watched it since.an outside tool that can do anything
  1. howYour IT provider's remote-management tool is broken into.
    what leavesevery computer it manageswho ends up with ita ransomware crew hitting all its clients at once
    it happenedKaseya VSA, 2021
  2. howThe master key of a remote-support product is stolen.
    what leavesstaff computers and the documents on themwho ends up with ita state-sponsored group
    it happenedUS Treasury and BeyondTrust, 2024
  3. howA support contractor's laptop is taken over for a short while.
    what leavesreach into customer accountswho ends up with itan extortion gang posting screenshots
    it happenedOkta / Lapsus$, 2022

4what works for a small company

Know which accounts they are. List every permanent Global Administrator or super admin and every account excluded from your MFA or Conditional Access policies. Each one is either a deliberate emergency account with a reason written down, or a gap. An emergency account is also different from a shared admin login used for daily work, and should never double as one.

Give it a credential that survives the rules. Two accounts, each with a passkey on two hardware keys, stored in two places. No personal phone, no personal recovery email. Microsoft recommends a passkey for these accounts and credentials kept in separate safes, and says a passkey satisfies the mandatory MFA requirement.

Make every sign-in loud. An alert on any sign-in by these accounts, sent to a phone someone answers. Microsoft documents this with Azure Monitor; in Google Workspace, admin email alerts cover the same ground.

Test it on a schedule. Microsoft's guidance says to validate the accounts at least every 90 days, after IT staff change, and when subscriptions change. Each test checks the sign-in, the alert, and that nobody added their own phone. The founder departure guide treats once a year as the floor; a calendar entry every quarter is easier to keep.

Review every use. Drill, real emergency or misuse: write down the date, who and why, and replace the key if it left the safe.

When did anyone last sign in to your emergency account, and who would have noticed?

on the record

← all topics

your turn

what you can do about it

  1. this week

    Sign in to your emergency admin account with the credential from the safe and confirm someone got the alert.

    An emergency account nobody has tested is a guess. The test shows whether it still signs in under today's MFA rules, and whether anyone would notice if a stranger used it.

  2. go deeper

  3. if you're not sure where to start

    Run the offboarding checklist on one leaver. It covers every account and token to close when someone leaves.

    open the checklist →

the letter · every two weeks

one access gap in your inbox, with the fix.

Every issue takes one way access creeps in, shows a real incident where it went wrong, and ends with one thing to check that week. Written for the people who run the tools, without a security team behind them.

  • 1one mechanism, explained plainly
  • 2one real incident, taken apart
  • 3one step you can take on Monday

get the next issue

One email every two weeks. Your address stays here and is never passed to vendors. One click to unsubscribe. Privacy.