On this page
If your Microsoft 365 tenant ever connected ConnectWise Cloud Backup (formerly SkyKick) or a ConnectWise Cloud Management app, disable that enterprise application now, even if nothing looks wrong. A vendor credential for the product was abused on 15 October 2026. Then read the service principal sign-in logs for the published application ID, and keep that old app off: ConnectWise now restores backups through a new application.
| Fact | Detail | Source |
|---|---|---|
| What was abused | A Microsoft 365 credential (API secret) belonging to ConnectWise Cloud Backup, formerly SkyKick | Huntress |
| Application ID | 506e08f9-6e9a-4da4-942b-a099a03644ae (Cloud Backup) | Huntress, ConnectWise |
| Main activity | 15 October 2026; Sophos puts the token sweep at about 96 minutes | Sophos |
| Confirmed scale | 162 tenants with unauthorized sign-ins as of 19:00 UTC on 16 October (Huntress telemetry) | Huntress |
| Later count | Nearly 1,000 tenants showed sign-in activity from the attacker IP, over 300 of them successful (Huntress, 17 October update) | Huntress |
| Vendor timeline | Incident response began 14 October 2026; advisory posted 16 October, 17:47 UTC | ConnectWise |
| Affected services | Cloud Backup; SaaS Security and Cloud Management suspended as a precaution | ConnectWise |
| Attribution | Storm-0940, a China state-aligned actor using the open-source ARTEX agentic AI framework. After taking control of ConnectWise app identities it attempted sign-ins to more than 2,250 tenants | Microsoft, 21 October |
| CVE | None assigned in the sources we checked | n/a |
| Vendor status | Phased restoration from 28 October through a new app, ConnectWise Cloud Backup for M365, which each tenant must re-authorize. Phases 1 and 2 (backups and restores, portal and monitoring) complete and phase 3 in progress at the 30 October, 21:00 UTC update | ConnectWise |
| Checked | 4 November 2026 | this page |
Who is affected, and who is not?
The affected population is MSPs and businesses that connected ConnectWise Cloud Backup or Cloud Management to a Microsoft 365 tenant. Those products were granted access to tenant data so they could back it up or manage it. If you never installed either, there is nothing in your tenant to disable.
The numbers need careful reading. Huntress confirmed unauthorized sign-ins in 162 tenants on 16 October. Its later count of nearly 1,000 tenants refers to sign-in activity from the attacker's address, of which over 300 succeeded. A sign-in is not the same as data taken, and neither vendor has published how many tenants lost data.
Two further points matter. Huntress confirmed abuse of additional ConnectWise Cloud Management apps (Exchange Online, MS Teams, SharePoint Online). ConnectWise suspended SaaS Security and Cloud Management as a precaution. If your tenant lists any of these, treat them the same way as Cloud Backup.
What is the confusion about "a vendor credential"?
Most owners picture a stolen password. This was a different kind of key. Cloud Backup is a multi-tenant application. Each customer's admin consents to it once, and it then holds standing permissions in that customer's tenant. Microsoft calls this app-only access: the app acts as itself, with no user signed in, and can reach any data its permission covers.
The app proves who it is with its own credential, a secret or certificate, using the client credentials flow. Whoever holds that secret can ask Microsoft for a token for any tenant that has consented. The vendor's one secret therefore sat in front of every customer's consent.
That has three consequences for you:
- Multi-factor sign-in and conditional access for your staff do not apply, because no staff member signs in.
- Your own password hygiene did not cause this and cannot undo it.
- Your only controls are the ones on the app's side of the door in your tenant: its consent, its roles, and the logs.
Sophos describes the main run as one app-only Graph token requested per targeted tenant, in ascending tenant ID order.
What did the attacker do inside tenants?
The two vendors saw different things, and both views are worth reading before you decide how deep to dig.
| Question | Huntress | Sophos |
|---|---|---|
| Roles changed? | Service principals were assigned directory roles in about 50 tenants from 01:58 UTC on 15 October, including Global Administrator | No evidence of new role assignments in its data |
| New credentials planted? | Credentials were added to two other applications in one tenant, including a security application | No evidence of new secrets or applications |
| Pace | In one tenant, role and credential changes within about five minutes, 03:44 to 03:49 UTC on 15 October | About 96 minutes of token requests |
| Scope of view | Its own protected tenants | The activity it observed |
The reasonable reading is that the actor did more in some tenants than in others, and each vendor reports its own telemetry. Neither report can speak for your tenant. Microsoft describes Global Administrator as able to manage all aspects of Entra ID and Microsoft services that use Entra identities, which is why role assignments are the first thing to check.
How do I fix it? (Disable first, investigate second)
Order matters. Disabling takes minutes and stops new tokens being issued for the app in your tenant. Investigation takes longer and depends on logs you may or may not have.
Step 1: disable the ConnectWise apps
Huntress says to disable the application now, even if you see nothing unusual, and to disable it rather than delete it. Microsoft documents the route: in the Entra admin center go to Enterprise apps, select the app, open Properties, and set Enabled for users to sign-in? to No. Microsoft states this prevents tokens being issued for that application in your tenant.
- Open Enterprise apps and search for ConnectWise, SkyKick, Cloud Backup and Cloud Management
- Set each one to not enabled for sign-in, and save
- Review the permissions each app holds, and revoke admin consent only if your provider agrees. Microsoft's permissions guide explains how to review and revoke
- Write down the time you did it
Step 2: read the logs
| Where | What to look for | Source |
|---|---|---|
| Entra admin center, Monitoring and health, Sign-in logs, Service principal sign-ins tab | Application ID 506e08f9-6e9a-4da4-942b-a099a03644ae; the address 144.172.64.128; activity outside Azure ranges | Huntress, ConnectWise |
| Entra audit logs | A ConnectWise or SkyKick app adding itself to a directory role, or changing other applications; compare user agents (vendor traffic showed kiota-dotnet, the suspicious changes Python-urllib/3.12) | Huntress |
| Entra audit logs, applications | New secrets, certificates or role assignments since 13 October | Huntress, Sophos (no such changes in its data) |
| Microsoft Graph activity logs, if you had collection on | Requests by that app's service principal on 15 to 17 October | Microsoft Learn |
Microsoft's docs explain why these exist. Service principal sign-ins record sign-ins by apps that present their own credential, and audit logs capture changes to applications and service principals. Graph activity logs need an Entra ID P1 or P2 licence and a diagnostic setting, so check whether yours was switched on before the incident.
The date window to cover is 13 to 17 October. Huntress saw password spraying from 13 October, the main sign-ins on 15 October and a second targeted wave on 17 October.
Step 3: if you are an MSP, repeat it per customer
We use ConnectWise Cloud Backup or Cloud Management on our managed tenants. For each customer tenant, please confirm in writing that the enterprise application was disabled, at what time, and whether the service principal sign-in and audit logs for 13 to 17 October show any activity from the published application ID or IP address.
Send that to whoever administers each tenant, or run the checks yourself under your management agreement. The app is a separate object in every tenant. Agencies that manage client sites as well as tenants will recognise the same one-key-many-clients risk in management plugins as a single point of failure, and the October incidents guide for agencies covers the wider month.
Step 4: do not re-enable early
Huntress says not to re-enable any ConnectWise application until ConnectWise confirms all credentials have been reset. ConnectWise says it completed a full reset of its accounts and credentials on 19 October, and on 28 October it began phased restoration, starting with Microsoft 365 backups and restores. The route back is a new application, ConnectWise Cloud Backup for M365: reset your partner password, complete MFA setup and re-authorize each tenant in it. Do not re-enable the old, disabled applications unless ConnectWise instructs you to.
At its 30 October update phases 1 and 2 (backups and restores, then the customer portal and monitoring) were complete, phase 3 (reporting, export and billing) was in progress and phase 4 (notifications) was still to come. ConnectWise adds that re-authorization alone does not confirm that backups have resumed, so check that a fresh backup job has completed for each tenant. Backups through this product stopped while the service was suspended, so record that gap. If you rely on it as your only copy of mailbox data, say so to your management today. When you do restore from any backup, choose a clean restore point first.
Am I affected? Where you stand right now
Your tenant's own logs are the only place this question gets answered. A scan of your website cannot see app consents, roles or sign-ins in Microsoft 365.
- Detected The service principal sign-in logs show the published application ID signing in from the published IP address, or the audit logs show role or credential changes you cannot explain. Treat the tenant as an incident. Disable the app, remove unexplained roles and credentials, and follow your incident process.
- Unknown You use a ConnectWise app but cannot read the logs for 13 to 17 October, or you are not sure which of your customers' tenants use it. This is the common case. Disable the app, then work out who can read the logs.
- Confirmed by owner evidence You disabled the app, reviewed the sign-in and audit logs for the whole window, and the roles and credentials in your tenant all match what you expect. That is evidence for your tenant, for the logs you have. Keep the old app off and restore backups through ConnectWise's new application.
No scan, ours included, moves you from "unknown" to "confirmed".
Was anything exploited?
Yes, in the sense that matters here: unauthorized access happened. Huntress confirmed unauthorized sign-ins in 162 tenants within a day of publishing, and ConnectWise's advisory describes unauthorized access to application identities in the Cloud Backup environment. What is not public is what the actor read or took in each tenant, and how many tenants lost data. Microsoft attributes the campaign to Storm-0940, a China state-aligned actor, which it says used the open-source ARTEX agentic AI framework and, after taking control of the identities of multiple ConnectWise cloud applications, attempted sign-ins to more than 2,250 tenants. Microsoft says most of that targeting was not successful. ConnectWise relayed the same attribution on 21 October and has not said how the initial access occurred.
Huntress also reported a second wave on 17 October against 16 tenants it protects, in policy, national security, trade and critical mineral sectors. That describes who was targeted among its own customers, not who was targeted overall.
What should I change so this hurts less next time?
This incident is a clean example of a trust you grant once and then forget. Use it to look at everything else with standing access to your tenant. Revoking connected-app OAuth tokens after the Klue to Salesforce incident follows the same logic.
| Question | Why it matters |
|---|---|
| Which enterprise apps hold application permissions (not delegated)? | They act with no user, so user-level controls do not apply |
| Which of those can read all mail, all files or the directory? | Those are the ones where one vendor secret becomes your breach |
| Who consented, and is the vendor still a supplier? | Old consents outlive contracts |
| Are service principal sign-ins and Graph activity logs being kept? | You cannot investigate what you did not record |
| Do you have a named owner for each vendor connection? | Someone has to act on the next advisory |
The same review applies to AI tools connected to your accounts, which we cover for Google in auditing AI tools connected to Google Workspace. If you do find unexplained credentials, the order for rotating exposed credentials is laid out step by step, and a step-by-step incident response plan covers evidence and communication, and the other breach write-ups show how similar incidents played out. Red's scans look at your public website from outside, which is a different surface from your Microsoft 365 tenant, as this piece on the limits of outside-in scanning explains.
Questions people ask
Was my Microsoft 365 tenant affected by the ConnectWise SkyKick breach?
Only a tenant that had one of the affected ConnectWise apps consented can be affected. Check Enterprise apps in the Entra admin center for Cloud Backup (formerly SkyKick) or any ConnectWise Cloud Management app, then read the service principal sign-in logs for the published application ID. No sign-ins from the published IP address means nothing was seen in the logs you have. It does not show that nothing happened.
Should I disable the ConnectWise apps even if I see nothing unusual?
Yes. Huntress recommends it, and ConnectWise tells customers not to re-enable the previously disabled applications unless it instructs them to. Disable the enterprise application rather than deleting it, so you keep its settings and logs. To restore backups, reset your partner password, set up MFA, re-authorize each tenant in the new ConnectWise Cloud Backup for M365 application, and confirm a fresh backup job completes.
Do I need to rotate my users' passwords?
The abused secret belonged to the vendor application, not to your users, so a password reset is not the main fix. The main steps are disabling the app, reviewing credentials and role assignments added to other applications in your tenant, and removing anything you cannot explain. If you find changes, treat the tenant as an incident and follow your response plan.
When can I turn Cloud Backup back on?
Not by switching the old app back on. From 28 October ConnectWise restores service through a new application, ConnectWise Cloud Backup for M365: partners reset their password, set up MFA and re-authorize each Microsoft 365 tenant in it. The old, disabled applications stay off unless ConnectWise instructs otherwise. At its 30 October update, backups, restores, the portal and monitoring were back, and ConnectWise notes that re-authorization alone does not confirm backups have resumed.
I am an MSP. Do I check each customer tenant?
Yes. ConnectWise's guidance asks partners to review their own Microsoft 365 tenant and each managed customer tenant. The enterprise application exists separately in every tenant, so disabling it in yours does nothing for theirs.
Can an outside scan tell me if this affected my tenant?
No. App consents, role assignments and sign-in logs live inside your Microsoft 365 tenant, and a scan of your website cannot see them. Only the tenant's own logs can answer it.
Sources
- ConnectWise Trust Center advisories: Cloud Backup service interruptionThe vendor's own statement: incident response began 14 October 2026, advisory posted 16 October at 17:47 UTC, affected services, customer actions, phased restoration from 28 October. Latest update read: 30 October, 21:00 UTC. Checked 4 November 2026.
- Huntress: ConnectWise (SkyKick) vendor credentials abused in M365 attacks (16 October 2026, updated through 27 October)Timeline, tenant counts, application IDs, log-review steps and the re-enable warning. The 27 October update adds Microsoft's attribution and no new counts. Checked 4 November 2026.
- Microsoft Threat Intelligence: Storm-0940 uses ARTEX in ConnectWise supply chain compromise (21 October 2026)Attribution to a China state-aligned actor, the ARTEX agentic AI framework, and sign-in attempts against more than 2,250 tenants. Read on 4 November 2026.
- Sophos: ConnectWise Cloud Backup application abused in token sweepAn independent view of the same 15 October activity, with a different read on what the actor changed. Checked 24 October 2026.
- Microsoft Learn: Review permissions granted to enterprise applications
- Microsoft Learn: Disable user sign-in for an application
- Microsoft Learn: Overview of permissions and consent
- Microsoft Learn: OAuth 2.0 client credentials flow
- Microsoft Learn: Service principal sign-in logs
- Microsoft Learn: Sign-in logs in Microsoft Entra ID
- Microsoft Learn: Audit logs in Microsoft Entra ID
- Microsoft Learn: Microsoft Graph activity logs
- Microsoft Learn: Microsoft Entra built-in roles reference
Facts on this page are dated and point at the primary source next to them. If a vendor status changes after the date shown, the vendor's page wins; tell us at support@run.cloud and we will update it.