← INTELLIGENCE HUB

One Password, Every Door (Twelve Ways In, Part 3)

NOVEMBER 19, 2026 · UMERCS

One Password, Every Door (Twelve Ways In, Part 3)

Most break-ins start with someone guessing. Not a clever exploit, not a zero-day, not a piece of malware with a name. A password. One that a real employee chose, tried against a real login page, at a rate slow enough to look like a busy Tuesday.

This is the third installment of Twelve Ways In, a series working through the twelve ways an attacker actually gets into a small or mid-size business, one path at a time. Part 2 covered the password that was already stolen. This one covers the password nobody stole at all - the attacker just guessed it, along with a few hundred other guesses, and one of them worked.

The technique has an unglamorous name: password spraying. It is the least sophisticated attack in this series and consistently one of the most effective, because it does not need a vulnerability. It needs a human resources policy, a login page, and patience.

The technique, in order

Spraying is brute force with the arithmetic reversed. A classic brute force attack takes one account and tries thousands of passwords against it. That fails for two reasons: it trips an account lockout within a minute, and it produces a wall of failed logins in a log that someone eventually reads.

Spraying flips it. The attacker takes one password and tries it against every account in the company, once. Then waits, then tries a second password, once each. Against three hundred mailboxes, that is three hundred attempts per round, spread across an hour, which is roughly five attempts a minute. No single account comes close to a lockout threshold, and the failed logins blend into the background noise of a Monday morning.

The attacker needs two things before starting. A list of usernames, which is easy - firstname.lastname at a company domain is the default format nearly everywhere, and LinkedIn supplies the names for free. And a short list of passwords people actually choose, which is easier, because it is the same short list every year. Season plus year. Company name plus exclamation mark. Welcome plus a number. The password the new hire was given in January and never changed.

The old internet joke is that the password is always 'hunter2.' In practice it is seasonal, capitalized, and ends in a digit, because the system made someone change it in October and that was the first thing that satisfied the rule.

Why it works: the rule that protects one account protects all of them

Account lockout is a per-account control. It counts failed attempts against one username and stops after five or ten. That is exactly the right defence against brute force, and it is nearly irrelevant against a spray, because the spray never fails against one account repeatedly. It fails once, everywhere, and moves on.

The other control people rely on is complexity. Twelve characters, mixed case, a number, a symbol. Rules like that produce predictable output, and attackers know the output. The pattern is well documented: when people are forced to add a digit and a symbol, they add 1 and !, and they put them at the end. The rule does not widen the search space nearly as much as the compliance form suggests.

Every organisation has a handful of accounts that make a spray worthwhile. A shared mailbox for invoices. A reception address. An account for the person who left in March and was never disabled. The spray does not need to compromise the whole company. It needs one mailbox with the right kind of email in it.

There is also a quieter reason spraying keeps working: it is cheap and unattributable. The infrastructure is a rented server and a script. When one hosting provider shuts the activity down, the attacker rents from another, and the target list comes with them.

The documented case: 81 million attempts, 78 mailboxes

In July 2026, Huntress published findings on an automated spray campaign against Microsoft's Azure command-line interface. Between June 12 and June 26, the operator made more than 81 million login attempts and compromised at least 78 Microsoft accounts across 64 organisations. The targeting was not industry-specific. It was driven entirely by which username and password combinations already existed on breach lists circulating in criminal markets.

The detail that matters is how it got past multi-factor authentication. Many of the affected organisations had Conditional Access policies enabled, and the campaign used a deprecated OAuth flow called Resource Owner Password Credentials, or ROPC, to authenticate. ROPC sends a username and password straight to the authorisation server in exchange for a token. Because it does not go through the normal interactive sign-in where Conditional Access is enforced, policies that looked healthy on a dashboard did not apply to those logins.

Policies failed in familiar, narrow ways: MFA enforced for specific applications but not all cloud applications, so Azure CLI was excluded. MFA enforced for administrators but not for everyone else. MFA enforced only outside trusted locations. Eight of the affected businesses had no MFA policy at all.

The volume is the part that should sit with you. Huntress measured a 155-fold increase in credential spray attacks across its customer base, with an average of about 1,964 failed attempts per month per protected tenant. Daily compromise counts stayed low - two to four accounts a day for most of the campaign - which is precisely what a spray is designed to produce. When one hosting provider suspended the activity, it moved to a second network, then a third, restoring volume to roughly 1.5 million attempts per day within weeks.

What one mailbox is worth

A compromised mailbox is not a trophy. It is a base of operations, and the attacker usually spends a day or two inside before doing anything loud.

The first moves are quiet. Forwarding rules that copy invoice-related mail to an outside address. Inbox rules that file anything from the finance director into a folder nobody opens. A sweep of the sent items and the contact list, which tells the attacker who approves payments and what the company's voice sounds like in writing.

The second move is internal phishing from a real address. A message from a colleague's actual mailbox, with the real signature and the real thread history, gets clicked at rates no external phishing campaign can approach. In Microsoft 365 environments, the same mailbox often grants access to Teams, SharePoint and the finance drive, so the attacker can read contracts, learn the payment run schedule, and send a change-of-bank-details request that arrives from an account the finance team has corresponded with for years.

None of this requires a vulnerability, an exploit, or a piece of malicious software on any endpoint. Which is why it does not generate the alerts most security tools are built to catch.

What you would actually see

Almost nothing. That is the defining feature. There is no outage, no ransomware note, no help desk ticket. Email keeps flowing. The website is up. The only evidence sits in authentication logs, in the form of failed sign-ins from unfamiliar addresses, most of which look exactly like the ordinary failures that happen when someone mistypes a password.

The signals that separate a spray from ordinary noise are specific. Many distinct accounts failing from a small set of addresses within a short window. Successful sign-ins for accounts that had a run of failures just before. Authentication against legacy protocols or older client types - ROPC, IMAP, POP, SMTP - that modern applications no longer use. A successful login from a country where the company has no staff, followed by mailbox rule creation within a few hours.

If you have never looked at your sign-in logs for legacy authentication, the exercise is worth an afternoon. What you are counting is how many of your successful logins last month used a protocol that cannot enforce MFA. In a lot of environments the answer surprises the person who asked.

The controls that close this door

Nothing on this list is exotic, and none of it requires new hardware. The gap is almost always configuration breadth rather than missing products.

Require MFA for all users, all cloud applications, and all client application types. The exclusions are where sprays get in: the app-specific scope that skips the command-line interface, the group scope that covers admins and nobody else, the trusted-location exemption that a remote attacker never trips.

Block legacy authentication outright. ROPC, IMAP, POP and basic SMTP cannot present an MFA challenge, so any policy that depends on MFA simply does not apply to them. Blocking them breaks older scanners and old phone clients, which is exactly the conversation to have once rather than leaving the door open.

Turn on a lockout or smart lockout policy and keep the threshold low enough to matter. Pair it with a banned password list that rejects the seasonal and company-name patterns before a user can set them. Microsoft and Google both publish these; the good ones also block close variants.

Alert on the spray pattern, not the individual failure: one address failing against many accounts, or one password failing across many accounts, inside a short window. That is a query, not a product.

Screen for compromised passwords at sign-in, and kill reuse across personal and work accounts. A stolen password from a personal site is the Part 2 problem, but it is also the reason a spray succeeds on the first guess: the credential was already valid.

And keep the inventory honest. The mailbox of the person who left in March, the shared inbox for invoices that five people can read, the service account with a password from 2019 and no MFA - these are the accounts a spray is looking for, and they are the ones that never appear on an org chart.

The five questions to ask your IT team

One: is MFA required for all users, all cloud apps, and all client app types, including command-line and legacy protocols? Two: is legacy authentication blocked, and can you show me the sign-in report that proves it? Three: what is our lockout threshold, and is there a banned password list? Four: do we alert when one address fails against many accounts, or when one password fails across many? Five: how many enabled accounts belong to people who no longer work here?

If the answers are uncertain, the fix is configuration and an afternoon, not a budget cycle. Spray attacks are the cheapest thing an attacker can do to your business, and the cheapest to close - which makes them a fair test of whether the security you were sold is actually switched on.

Sources

  • The Hacker News, Azure CLI Password Spray Hits at Least 78 Microsoft Accounts in 81M+ Attempts (reporting Huntress research; campaign June 12-26 2026, ROPC bypass of Conditional Access, 155x surge, ASN migration), July 1 2026, updated July 21 2026: https://thehackernews.com/2026/07/azure-cli-password-spray-hits-at-least.html
  • SecurityWeek, Massive Password Spray Campaign Targeting Azure CLI, July 1 2026: https://www.securityweek.com/massive-password-spray-campaign-targeting-azure-cli/
  • Microsoft Learn, Password protection and smart lockout in Microsoft Entra ID (lockout thresholds, custom banned password lists): https://learn.microsoft.com/en-us/entra/identity/authentication/howto-password-smart-lockout
  • Microsoft Learn, Block legacy authentication with Conditional Access (legacy protocols cannot enforce MFA): https://learn.microsoft.com/en-us/entra/identity/conditional-access/block-legacy-authentication

STEP 1/6