dogesec Logo
Did an AI Agent Make Itself Global Admin? Inside the ConnectWise Hack

Did an AI Agent Make Itself Global Admin? Inside the ConnectWise Hack

Microsoft says Storm-0940 used ConnectWise's app identities and an AI pentest tool to try 2,250+ Microsoft 365 tenants. The evidence as STIX, plus a Sigma rule.

Category: Research

tl;dr

  • On 15 October 2026, attackers holding the Microsoft 365 app identities behind ConnectWise Cloud Backup tried to sign in to more than 2,250 customer tenants. Nobody has said how many were compromised.
  • Microsoft blames Storm-0940, a group it calls China-state aligned, and says the attack ran through ARTEX, an open-source AI penetration-testing framework. China’s embassy in Washington rejected the accusation, Reuters reported.
  • ConnectWise switched off seven app identities on 17 October, doesn’t know how its credentials were taken, and was still restoring the service on 4 November.

Five minutes to Global Admin

At 01:58 UTC on 15 October 2026, someone started signing in to ConnectWise customers’ Microsoft 365 tenants as ConnectWise’s own apps. In one tenant that Huntress, a security vendor that monitors tenants for managed service providers (MSPs), watches, the ConnectWise Teams app added itself to Global Administrator, the role that can do anything in a tenant, between 03:44 and 03:49. A Python script then added client secrets, valid until 2028, to an MSP’s app and to Huntress’s own app. Nobody phished a user or cracked a password. As far as Microsoft 365 could tell, this was ConnectWise.

That works because of how SaaS vendors plug into Microsoft 365. ConnectWise Cloud Backup, formerly SkyKick, backs up Microsoft 365 for MSPs and their customers. Each customer grants ConnectWise’s app its permissions once, by admin consent, and Entra ID, Microsoft’s identity service, creates a service principal, a local copy of the app, in their tenant. The app then signs in as itself with a secret or certificate that ConnectWise holds: no user, no MFA prompt, nothing in the user sign-in logs. The same app ID works in every tenant that consented, so one secret reaches every customer.

Microsoft says the people holding that secret were Storm-0940, and that an AI agent did much of the work. Whether an agent really chose to make a backup vendor’s Teams app Global Admin is what the rest of the evidence has to answer.

Seven apps switched off

ConnectWise’s first advisory went out at 17:47 UTC on 16 October, naming 144[.]172[.]64[.]128 and the user agents Python-urllib/3.12 and python-requests/2.31. At 05:01 the next morning it cut Cloud Backup’s access to customer tenants, and at 07:00 it disabled seven app identities, from Cloud Backup to SharePoint Online. It revoked their certificates that afternoon, added 144[.]172[.]64[.]99 and 144[.]172[.]64[.]60 on 18 October, and reset every credential on 19 October.

Phased restoration was under way by 22 October. On 28 October ConnectWise moved partners to a new app, ConnectWise Cloud Backup for M365, and backups resumed for each tenant re-authorised with it. On 4 November it said restoration “continues”. What it still can’t tell partners is whether any specific tenant was compromised, or how the credentials were obtained (FAQs 7 and 9 of the advisory).

73 posts and five different numbers

By then the story was everywhere. From 1 August to 13 November, Obstracts’ feeds carried 73 posts on the story: 19 on the ConnectWise compromise, 47 on ARTEX more widely and 7 on neighbouring stories. Depending on which you read, the attackers reached 162, 173, more than 300, about 1,000 or more than 2,250 tenants.

They measure different things. Huntress counted 162 tenants with confirmed unauthorised sign-ins on 16 October, then about 1,000 with sign-in activity and more than 300 with a successful sign-in. The Hacker News reported 173, a figure no longer on Huntress’s page. Microsoft’s 2,250 counts tenants where sign-ins were attempted, and Microsoft says most of the targeting failed. And a sign-in isn’t a compromise: by 19 October Huntress had found post-compromise activity in two tenants. Nobody has published how many were actually compromised.

The primary sources:

Source What it saw What it lacks
ConnectWise advisories (16 Oct to 4 Nov) The victim’s account: addresses, app identities, audit operations to review, restoration How its credentials were taken; which tenants were compromised
Huntress (16 Oct) and Huntress (26 Oct) Minute-by-minute activity in the tenants it monitors, then whether an AI was in the loop Anything outside its own customers; attribution
Microsoft (21 Oct) Platform telemetry: the spray, the 2,250+ attempted tenants, the attribution, IoCs, an ATT&CK mapping Detailed evidence for the China link
Wiz (22 Oct) ARTEX’s origin, and Storm-0940’s interest in IT and software vendors Any mention of ConnectWise
FBI FLASH-20261113-001 (13 Nov) ARTEX abuse in general, with behavioural indicators Its own attribution of the campaign
CrowdStrike (7 Oct) and Recorded Future (16 Oct) ARTEX used against South Korean banks Any link to ConnectWise or Storm-0940

They agree on the timeline and the addresses. At the edges, Huntress calls the stolen credential both an “API secret” and a “client assertion”, Microsoft counts four application credentials where ConnectWise disabled seven apps, and only Microsoft names who did it.

Who Microsoft blames

On 21 October Microsoft said it “assesses” the attack as the work of Storm-0940, a China-state aligned group active since 2021 with no known aliases. It gave two reasons:

  • the same server, 144[.]172[.]64[.]99, ran a password spray from 10 to 14 October and then took part in the ConnectWise activity;
  • Storm-0940 has a history of password spraying Microsoft 365 accounts.

It gave no detailed evidence for the China link. ConnectWise repeated the attribution, citing Microsoft. Wiz described Storm-0940 using ARTEX against IT and software vendors the next day, without naming ConnectWise. The FBI describes the campaign, citing Microsoft, but names neither Storm-0940 nor China. A spokesperson for China’s embassy in Washington said China “opposes and fights hacking activities in accordance with the law”, Reuters reported.

In 2024 Microsoft described Storm-0940 as a patient sprayer, working through a covert network it calls CovertNetwork-1658 (also known as Quad7) at about one attempt per account per day in about 80% of cases. In October 2026 the same group, if Microsoft is right, made more than 151,000 failed sign-ins against 327 accounts, peaking at about 2,900 a minute, and got nothing. The FBI calls ARTEX activity “forensically noisy”. Storm-0940 traded stealth for speed, and a fast attacker leaves a loud audit log.

What ARTEX is

The other half of Microsoft’s claim is the tool. ARTEX is a multi-agent penetration-testing framework: planner and worker agents run ordinary tools such as nmap, curl and Impacket, according to Wiz. Microsoft adds that it is written in Go and stores what its agents find in a graph. It has no model of its own; it runs on a large language model (LLM) of the operator’s choice.

A Chinese developer, named by the FBI as “Autumn-27”, published it on GitHub in July 2026. It won Baidu’s Agent+ red-team challenge in September and went closed-source on 8 October, after more than 12,000 forks; forks, mirrors and an “ARTEX PRO” remain. The FBI says it has been abused since at least August and counts more than 1,600 exposed consoles.

ConnectWise wasn’t its first victim. On 7 October CrowdStrike described ARTEX being used against South Korean finance by an “unknown threat actor”, assessed with moderate confidence as Chinese-speaking and financially motivated. Recorded Future calls the actor TAG-217, traces its infrastructure to 17 August, and cites reports of at least seven institutions and 68,000 people; the FBI says about 67,000. Neither ties it to Storm-0940, and the FBI lists it as a separate campaign: the two cases share a tool and nothing more.

Was an AI really driving?

Huntress went back through its logs on 26 October in a post called AI, Scripts, Operator, and the title is its answer.

Some of it looks like reasoning about failure. A certificate upload failed at 03:48:24 and the run ended without retrying. The next run, 55 seconds later, retried the same error at 3.5 and 7.1 seconds until it worked, and carried that retry into a second tenant. Tools changed in under a minute, user agents looked hand-set and credentials were changed by trial and error. Some of it looks like a plain loop: token requests walked about 1,000 tenant IDs in ascending order, and requests against the Teams service principal kept firing for 22 minutes after it appears to have been deleted, at about 03:57. Bursts of 100 failed requests in 10 seconds fit either, and the FBI lists “compressed bursts” among ARTEX’s tells.

Huntress rates the likelihood of an LLM in the loop as low-to-medium on its own telemetry, and medium-to-high once Microsoft’s attribution is added. So did an AI agent make itself Global Admin? A ConnectWise app certainly did. Whether an agent chose that step, a script ran it or an operator typed it is a probability, and the logs can’t settle it.

Claims that don’t hold up

Some of the 73 posts outran the evidence. A securityonline.info headline had attackers “breach over 300” tenants, when 300 was Huntress’s count of successful sign-ins. Some coverage, such as GovInfoSecurity’s, can be read as tying Storm-0940 to the Korean banks, when Microsoft only says public reporting had noted ARTEX there.

How the attack worked

In the order Microsoft and Huntress describe it, the attack has nine steps. The graph starts at step 2, because the spray isn’t known to have led to it. If Attack Flows are new to you, Beyond the ATT&CK Matrix: How to Build Dynamic Attack Flows with STIX explains how to read one.

  1. From 10 to 14 October, a password spray from 144[.]172[.]64[.]99 gets nowhere (Password Spraying, T1110.003; Microsoft lists Active Scanning, T1595, for the reconnaissance). It isn’t known to have led to step 2; Microsoft cites it as attribution evidence. Huntress saw curl/7.88.1 spraying four law-firm tenants from 144[.]172[.]64[.]128.
  2. Before 15 October, the attacker obtains credentials for ConnectWise’s apps. Microsoft maps this to Steal Application Access Token, T1528, but nobody has said how it happened.
  3. From 01:58 UTC on 15 October, it signs in to customer tenants as ConnectWise’s apps (Trusted Relationship, T1199; Microsoft also gives Supply Chain Compromise, T1195).
  4. The apps assign themselves Exchange Administrator and, for the Teams app, Global Administrator (Additional Cloud Roles, T1098.003). The Exchange Administrator assignments carried kiota-dotnet, which ConnectWise lists as expected vendor activity, so whether they were the attacker’s is unconfirmed.
  5. A script adds client secrets to other apps, access that would outlive ConnectWise’s own credentials (Additional Cloud Credentials, T1098.001).
  6. From 06:13 to 08:34 UTC, token requests from 144[.]172[.]64[.]128 sweep about 1,000 tenants; Microsoft counts more than 2,250 attempted, including over 200 sign-ins in under four minutes (Cloud Accounts, T1078.004).
  7. Accounts and apps are listed through Microsoft Graph, the API behind Microsoft 365 (Cloud Account, T1087.004).
  8. Reachable Azure resources are discovered (Cloud Infrastructure Discovery, T1580).
  9. Azure Key Vault secrets are read in bulk (Cloud Secrets Management Stores, T1555.006).

Microsoft doesn’t time steps 7 to 9, so their place after the sweep is likely rather than confirmed. Steps 4 and 5 matter most: a vendor’s app reaching for more than it was ever given, which leaves the clearest trace in the customer’s audit log.

What nobody has confirmed

The biggest gap is step 2. ConnectWise says how the credentials were obtained “remains under investigation”, and neither Microsoft nor Huntress says, beyond Microsoft ruling out a Microsoft vulnerability. Huntress’s two descriptions point different ways: an “API secret” sounds like a client secret copied from somewhere, while a “client assertion” is a signed token, made with a certificate’s private key or issued to one of the vendor’s own Azure services. Huntress says the credential “lives in the vendor’s own tenant”, and traced the access behind one of the credential additions to private Azure IPv6 space that ConnectWise gives as its apps’ normal egress.

What it needed is clear: access to wherever ConnectWise kept or used the credentials, and customers who had consented to its apps. Only a root cause from ConnectWise would settle the rest. If you want to model a step like this yourself, TTPs Are Missing the P: Lets Fix That explains how to record what an attacker did when no ATT&CK technique fits.

Three more questions stay open. How many tenants were compromised: two is a floor, and it comes from Huntress’s customers alone. What was taken: Graph reads aren’t written to the Entra audit log, so a tenant can know an app was there without knowing what it read. And how much of the work was an AI’s, which Huntress can only give as a probability.

What defenders can hunt for

The addresses are the easy part, and the weakest. 144[.]172[.]64[.]128, .99 and .60 are confirmed by ConnectWise and Microsoft, and Huntress puts the attacker on RouterHosting (AS14956), a hosting provider, so expect the addresses to change hands. Huntress adds 144[.]172[.]94[.]243 from the spray and 194[.]165[.]16[.]21 as only a possible link. Microsoft’s three VPN exits are short-lived and shared, and the user agents are low fidelity on their own. The ARTEX server 38[.]244[.]50[.]120 comes from CrowdStrike’s Korean case, not this one, so treat a hit as an ARTEX lead rather than a ConnectWise one. The graph below shows the IP addresses only; the user agents and app IDs are listed in the reference section at the end.

The behaviour is better. Huntress, Microsoft and ConnectWise all describe steps 4 and 5, and ConnectWise told partners which Entra audit operations to review, the ones the rule below matches.

A Sigma rule is a YAML file that detection tools convert into SIEM queries. If you are new to Sigma, Writing Effective Sigma Detection Rules: A Guide for Novice Detection Engineers explains each field. This one matches two ways: the seven ConnectWise app IDs, which should have been silent since 17 October, and any service principal doing one of those things to itself, as ConnectWise’s apps did in the tenants Huntress watched. The app ID is the same in every tenant, unlike the service principal ID, so one rule covers every customer of a vendor.

title: Vendor App Grants Itself a Directory Role or New Credentials
id: 3b8e2f4a-6c1d-4e9b-a7f2-9d0c5e8b1a64
status: experimental
description: Detects an Entra ID application, acting without a signed-in user, that adds a member to a directory role, adds credentials to an application or service principal, adds an owner or grants permissions, where the application is one of the seven ConnectWise app identities abused on 15 October 2026, or where the service principal is acting on itself. In the ConnectWise Cloud Backup compromise, ConnectWise service principals assigned themselves Exchange Administrator and Global Administrator, and a script then added client secrets to other applications.
references:
    - https://www.connectwise.com/company/trust/advisories
    - https://www.huntress.com/blog/vendor-credential-enables-unauthorized-access
    - https://www.huntress.com/blog/connectwise-attack-ai
    - https://www.microsoft.com/en-us/security/blog/2026/10/21/storm-0940-uses-artex-agentic-ai-connectwise-supply-chain-compromise/
author: dogesec
date: 2026-11-13
tags:
    - attack.privilege-escalation
    - attack.persistence
    - attack.t1098.003
    - attack.t1098.001
logsource:
    product: azure
    service: auditlogs
detection:
    selection_operation:
        OperationName:
            - 'Add member to role'
            - 'Add service principal credentials'
            - 'Add owner to application'
            - 'Add app role assignment to service principal'
            - 'Add delegated permission grant'
            - 'Consent to application'
    selection_operation_secrets:
        OperationName|contains: 'Certificates and secrets management'
    selection_connectwise:
        InitiatedBy.app.appId:
            - '506e08f9-6e9a-4da4-942b-a099a03644ae'  # Cloud Backup
            - '2399d300-6fca-4e6a-a686-399fefe75b83'  # Cloud Management
            - '24205314-2b89-4e07-a2a4-579e898a5640'  # MS Teams
            - 'b78f6a77-d4e5-46fb-87e5-d1648bd02519'  # Exchange Online
            - '0de14c77-b9b0-4137-9360-34878a929827'  # MS Graph
            - '412dcf96-176d-4b3a-a1b0-4981ad66971e'  # Partner Center
            - 'b2b55f5b-7758-4719-8d8f-88ae172e6bdb'  # SharePoint Online
    selection_self:
        InitiatedBy.app.servicePrincipalId|fieldref: TargetResources.id
    condition: (selection_operation or selection_operation_secrets) and (selection_connectwise or selection_self)
falsepositives:
    - An automation app that rotates its own certificate or secret through Microsoft Graph matches the self-action selection on 'Add service principal credentials'; list those apps' IDs in a filter
    - Legitimate ConnectWise automation before 17 October 2026, when the seven app IDs were in normal use; ConnectWise lists kiota-dotnet from Azure private IPv6 addresses as expected vendor activity, so in historical hunts check each match against that before treating it as the attacker's
level: high

The self-action selection compares the initiating service principal with the target, using Sigma’s fieldref modifier. Check that InitiatedBy.app.appId is populated in your own logs first: Microsoft’s Sentinel rule New access credential added to Application or Service Principal reads the initiating app’s displayName and servicePrincipalId, not appId. ConnectWise’s replacement app has its own ID and isn’t in the list.

The rule has blind spots. Adding secrets to another app isn’t a self-action, so the rule catches step 5 only through ConnectWise’s seven IDs, and only if, as Huntress implies, the script ran as the Teams app; a different vendor’s hijacked app doing the same would slip past. TargetResources is a list whose order depends on the operation, so test the self-action match on your own logs before trusting it.

App-only audit events carry no source IP, so the service principal sign-in logs are where you tie activity to an address. The Azure IPv6 address Huntress traced is shared with ConnectWise’s legitimate traffic, so it doesn’t identify the operator. Steps 7 to 9 never reach the Entra audit log; look for them in Graph activity logs, Azure Activity logs and Key Vault diagnostic logs. The same gap sank defenders at Binance: The $6.34 Billion Job Interview: How North Korea Robbed Binance shows an identity chain split across logs nobody read together.

The whole case in one graph

Put together, this is the whole case: the nine reports, the attack and Storm-0940, the group Microsoft blames, ARTEX and the separate Korean case, the nine steps, six of the IP addresses and the rule. If you are new to STIX, Understanding STIX 2.1 Objects explains what each node in the graph is.

A hunter gets one behaviour to watch, a vendor app granting itself roles or keys, and seven app IDs that should never appear again. An analyst gets the state of the case: Microsoft alone blames Storm-0940, the Korean bank thefts share a tool and nothing more, and the way in, the number of tenants lost and the AI’s share of the work are all still unknown.

In summary

ConnectWise’s backup apps were trusted in thousands of Microsoft 365 tenants, and on 15 October someone holding their credentials used that trust to try more than 2,250 of them. Microsoft blames Storm-0940 and ARTEX; Huntress’s logs show a mix of reasoning, scripting and human hands, and nobody can yet say how the credentials were stolen or how many tenants were lost. Whether or not an agent chose to make that Teams app Global Admin, the change landed in each customer’s audit log, and a rule that watches for vendor apps granting themselves roles or keys is built to catch exactly that.

Reference: IoCs and ATT&CK techniques

IoCs

Type Value What it is Linked to Confidence STIX ID
IPv4 144[.]172[.]64[.]128 Sign-ins and Graph token requests on 15 Oct; in ConnectWise’s first advisory ConnectWise Cloud Backup app identity abuse, October 2026 85 indicator--aaf7a4e4-8e7d-56d3-b3b4-6fd2645e5d1e
IPv4 144[.]172[.]64[.]99 Password spray 10-14 Oct, then ConnectWise activity ConnectWise Cloud Backup app identity abuse, October 2026 85 indicator--c3ee14aa-19f4-52b3-ade6-1b30a2b142bc
IPv4 144[.]172[.]64[.]60 Added by ConnectWise on 18 Oct; in Microsoft’s IoCs ConnectWise Cloud Backup app identity abuse, October 2026 85 indicator--aedb42d2-cb29-5167-8401-0604e0b67463
IPv4 144[.]172[.]94[.]243 Extra password-spray address (Huntress) ConnectWise Cloud Backup app identity abuse, October 2026 70 indicator--7dfc31ff-a941-5405-b545-1067554ec84a
IPv4 187[.]15[.]88[.]186 VPN exit address (Microsoft) ConnectWise Cloud Backup app identity abuse, October 2026 50 indicator--df7dd001-e069-50a2-b550-469644659867
IPv4 187[.]15[.]90[.]217 VPN exit address (Microsoft) ConnectWise Cloud Backup app identity abuse, October 2026 50 indicator--0e64d224-bb5f-5fe8-bc94-581d83656b6c
IPv4 187[.]15[.]89[.]186 VPN exit address (Microsoft) ConnectWise Cloud Backup app identity abuse, October 2026 50 indicator--121c790c-b076-59d9-840e-46285e802247
IPv4 194[.]165[.]16[.]21 Possible extra address; not confirmed (Huntress) ConnectWise Cloud Backup app identity abuse, October 2026 15 indicator--381ca911-da6f-5070-8b9a-e39d20025b61
User agent curl/7.88.1 Password spray ConnectWise Cloud Backup app identity abuse, October 2026 50 indicator--36ef6cba-18a5-5d58-8853-522bcbb79d5c
User agent Python-urllib/3.12 Tenant activity (ConnectWise) ConnectWise Cloud Backup app identity abuse, October 2026 50 indicator--c9fbdd1a-5de4-5319-b421-47511407c200
User agent python-requests/2.31.0 Tenant activity (ConnectWise writes python-requests/2.31) ConnectWise Cloud Backup app identity abuse, October 2026 50 indicator--160c012d-b19e-5cfb-8c6f-0fbfd0b3f628
User agent python-requests/2.34.2 Tenant activity (Microsoft) ConnectWise Cloud Backup app identity abuse, October 2026 50 indicator--0ee7f863-322e-584b-a996-fd2a59c70b4b
User agent azure-toolkit/1.0 Tenant activity (Microsoft) ConnectWise Cloud Backup app identity abuse, October 2026 50 indicator--b1b29318-e956-57d3-a380-5a17d0db4b0b
Entra app ID 506e08f9-6e9a-4da4-942b-a099a03644ae ConnectWise Cloud Backup app identity, disabled 17 Oct ConnectWise Cloud Backup app identity abuse, October 2026 85 indicator--d43f859e-c1b9-595c-97a8-cee53d960ea2
Entra app ID 2399d300-6fca-4e6a-a686-399fefe75b83 ConnectWise Cloud Management app identity, disabled 17 Oct ConnectWise Cloud Backup app identity abuse, October 2026 85 indicator--96d806fa-0dff-5429-be43-84b362a3e6af
Entra app ID 24205314-2b89-4e07-a2a4-579e898a5640 ConnectWise MS Teams app identity, disabled 17 Oct ConnectWise Cloud Backup app identity abuse, October 2026 85 indicator--cd64ecc9-11f7-5c0b-a3e9-605bcebd57c3
Entra app ID b78f6a77-d4e5-46fb-87e5-d1648bd02519 ConnectWise Exchange Online app identity, disabled 17 Oct ConnectWise Cloud Backup app identity abuse, October 2026 85 indicator--747dda3a-15cd-5698-8450-22971c5ef60d
Entra app ID 0de14c77-b9b0-4137-9360-34878a929827 ConnectWise MS Graph app identity, disabled 17 Oct ConnectWise Cloud Backup app identity abuse, October 2026 85 indicator--30d7a075-5121-509e-89b2-679fa95d02c9
Entra app ID 412dcf96-176d-4b3a-a1b0-4981ad66971e ConnectWise Partner Center app identity, disabled 17 Oct ConnectWise Cloud Backup app identity abuse, October 2026 85 indicator--f6caec25-b762-5d23-8d54-168a90f75367
Entra app ID b2b55f5b-7758-4719-8d8f-88ae172e6bdb ConnectWise SharePoint Online app identity, disabled 17 Oct ConnectWise Cloud Backup app identity abuse, October 2026 85 indicator--4457a540-9f95-5499-a4ba-b86c1a674340
IPv4 38[.]244[.]50[.]120 ARTEX server in the South Korean bank intrusions; not confirmed for this attack (CrowdStrike) ARTEX 50 indicator--96856ed6-dbc8-55ed-b854-def2c8746565

Techniques (Attack Flow)

Step Action Technique Tactic STIX ID
1 Password spray against Microsoft 365 accounts T1110.003 Password Spraying Credential Access attack-pattern--692074ae-bb62-4a5e-a735-02cb6bde458c
2 ConnectWise app credentials obtained (method unknown) None (procedure: Theft of a SaaS vendor’s multi-tenant app credentials; Microsoft maps it to T1528) Credential Access procedure--1b02b749-b196-51a7-b2b4-35334d1536f7
3 Sign-in to customer tenants as ConnectWise’s apps T1199 Trusted Relationship Initial Access attack-pattern--9fa07bef-9c81-421e-a8e5-ad4366c5a925
4 Apps assign themselves directory roles T1098.003 Additional Cloud Roles Privilege Escalation attack-pattern--2dbbdcd5-92cf-44c0-aea2-fe24783a6bc3
5 Client secrets added to other apps T1098.001 Additional Cloud Credentials Persistence attack-pattern--8a2f40cf-8325-47f9-96e4-b1ca4c7389bd
6 Token requests swept across customer tenants T1078.004 Cloud Accounts Initial Access attack-pattern--f232fa7a-025c-4d43-abc7-318e81a73d65
7 Accounts and apps listed through Microsoft Graph T1087.004 Cloud Account Discovery attack-pattern--8f104855-e5b7-4077-b1f5-bc3103b41abe
8 Azure resources discovered T1580 Cloud Infrastructure Discovery Discovery attack-pattern--57a3d31a-d04f-4663-b2da-7df8ec3f8c9d
9 Key Vault secrets read in bulk T1555.006 Cloud Secrets Management Stores Credential Access attack-pattern--cfb525cc-5494-401d-a82b-2539ca46a561

The Sigma rule is indicator--3b8e2f4a-6c1d-4e9b-a7f2-9d0c5e8b1a64, and the campaign is campaign--dc255787-6c47-56dd-a0f6-ea09bffb3ab5.

Posted by:

Dylan Thomas

Dylan Thomas, Do Only Good Everyday

Get new posts by email

New threat research and tutorials from dogesec, plus the weekly Obstracts threat brief. No spam, unsubscribe any time.

Thanks. Check your inbox to confirm your subscription.