mythsthe appsStorm-2372 device code phishing2372government, NGOs, IT, defense, health / Europe, North America, Africa, Middle East

everyone assumed:

if the sign-in page is really Microsoft's, it is safe to sign in

what was actually true:

From August 2024, a group Microsoft tracks as Storm-2372 got into government, NGO and company mailboxes by asking people to type a meeting code on Microsoft's genuine sign-in page.

how it happened

  1. whoa trusted-looking contactrapport built over WhatsApp, Signal or Teams
  2. the gapa code typed on Microsoft's pagethe victim signs in for the attacker's device
  3. thenmailbox searched"password", "admin", "credentials"
  4. the resultmail taken, colleagues phishedvictim count never published
the gap that let it happen the result

what everyone assumed

None of the targets has published what they believed, so this belief is an inference from how the lure worked. Most phishing advice comes down to one check: look at the address before you sign in. People who had learned it well were exactly the ones this campaign was built for. The page they landed on was Microsoft’s, the sign-in was theirs, and the prompt came to their own phone. Every visible check passed, which is why careful staff at ministries, NGOs and defense firms went ahead.

what happened

Microsoft Threat Intelligence reported on 13 February 2025 that it had found an active and successful device code phishing campaign by an actor it calls Storm-2372, running since August 2024. In Microsoft’s account, the targets were governments, NGOs, IT services, defense, telecommunications, health, higher education and energy organizations in Europe, North America, Africa and the Middle East.

The approach was patient. Microsoft says the actor most likely made first contact over WhatsApp, Signal or Teams, posing as a prominent person relevant to the target and building rapport before sending anything. Then came an invitation to an online event or meeting, usually a fake Teams invite by email, with a code presented as the meeting ID. Typing that code on Microsoft’s device sign-in page signed the target in for a device the attacker held. The device code guide explains why that hands over tokens and why MFA does not stop it.

Microsoft describes what the actor did with the access:

  • Searched the mailbox. Using Microsoft Graph, it looked for words such as “password”, “admin”, “teamviewer”, “anydesk”, “credentials”, “ministry” and “gov”, and sent the matching messages out.
  • Phished from inside. It sent new device code lures to colleagues from the compromised account, where a trusted sender makes the next invite more convincing.
  • Registered its own device. In an update dated 14 February 2025, Microsoft said that over the previous day the actor had switched to the client ID of the Microsoft Authentication Broker. That let it register an attacker-controlled device in the victim’s Entra ID, obtain a Primary Refresh Token and collect mail through that device, using proxies located near its victims to make sign-ins look ordinary.

Volexity, a security vendor, published its own findings the same day. It described several clusters it assessed as Russian, running device code campaigns from mid-January 2025, with lures posing as the US Department of State, Ukraine’s Ministry of Defence, the European Parliament and research institutions, over Signal, Element, email and Teams. Volexity noted that the codes are “only valid for 15 minutes once they are created”, which is why the attackers chatted with targets in real time and, in one case, used a page that generated a fresh code on each visit. Volexity does not say its clusters are the same group Microsoft calls Storm-2372, and this page does not treat them as one.

step by step

  1. August 2024the campaign startsMicrosoft later dates Storm-2372's activity to August 2024, against governments, NGOs and several industries.
  2. a meeting code typed on a real pageThe target enters the code on Microsoft's own sign-in page, and the session goes to the attacker's device.the assumption breaks here
  3. January 2025Volexity finds a breachVolexity traces a suspicious sign-in at one of its customers to a device code lure.
  4. 13 and 14 February 2025both write-ups go publicMicrosoft and Volexity publish the same day; a day later Microsoft reports the actor registering its own devices.
  5. 31 July 2026Microsoft updates the attributionStorm-2372 is assessed as an initial access sub-cluster of Midnight Blizzard.

how it surfaced

Volexity’s discovery began at the end of January 2025, when it saw a suspicious sign-in to a customer’s Microsoft 365 account, followed by rapid file downloads from OneDrive coming from rented servers and the Tor network. Microsoft has not described how it found its campaign. Both reports went public on 13 February 2025, and Microsoft’s update the next day showed the actor changing method within a day of being described.

Neither source gives a count of victims or names a victim organization. Volexity’s view, as vendor research based on what it saw at its own customers, was that the method had compromised accounts more successfully than most targeted spear phishing. Its explanation is short: there is no malicious link or attachment for a filter to catch.

Attribution moved over time. In February 2025 Microsoft said with moderate confidence that Storm-2372 aligns with Russian interests, and Volexity linked one of its clusters to APT29 at medium confidence. On 31 July 2026 Microsoft updated the same post to say it assessed Storm-2372 as “an initial access operations sub-cluster of Midnight Blizzard”, the group in the retired test tenant record. Both are assessments by the companies that published them.

Both reports gave the same main fix: block device code sign-in where it is not needed, and limit it with Conditional Access where it is. Volexity said the block stopped the attack in its tests. Microsoft added revoking the user’s refresh tokens after a compromise. Why a session can outlive a password change is the problem the token guide covers. Within months the method spread beyond state actors, which the device code guide follows through 2026.

A genuine sign-in page proves where you are typing. It says nothing about whose device you are signing in.

Ask who on your staff would type a meeting code sent by a contact they trust, and whether your tenant would let that sign-in through.

picture it at a company like yours

The company here is made up. The other two paths are not: they happened, and each links to the story.

your company
a normal dayA partner you have been chatting with sends a Teams invite, and the join page asks for a short code.
the gap nobody seesThe page is Microsoft's own, so nobody's training says anything is wrong.a second factor that can be relayed
  1. in this story
    the day it mattersThe meeting never starts, and someone else's laptop now reads that person's mailbox.
    what it costs youYour mail is searched for passwords and admin logins, and colleagues start getting invites from a trusted address.
  2. or it goes like this
    howA fake login page passes the password and code to the real site as they are typed.
    what leavesinternal documents, code and dashboardswho ends up with itthe attacker behind the page
    it happenedReddit, 2023
  3. or it goes like this
    howLogin prompts keep coming until a tired employee taps approve.
    what leavesadmin tools and company chatwho ends up with itan attacker who bought a contractor's password
    it happenedUber, 2022

sources

  1. Microsoft Threat Intelligence: Storm-2372 conducts device code phishing campaign, 13 February 2025 (updates 14 February 2025 and 31 July 2026)
  2. Volexity: Multiple Russian threat actors targeting Microsoft device code authentication, 13 February 2025

← all myths

your turn

what you can do about it

  1. this week

    Pull 30 days of device code sign-ins from the Entra logs and match each one to a known device.

    In the Entra sign-in logs, filter the authentication protocol to device code for the last 30 days. Every entry should map to a room screen, a CLI tool or another device someone can name. An entry nobody can explain is what Volexity's first finding looked like, so revoke that user's sessions, check their registered devices and mailbox rules, and then decide whether to block the flow for everyone else.

  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.