A mobile device management platform can lock, wipe, reconfigure and push software to every enrolled device at once, which makes it one of the most powerful and least protected systems in most organisations. The 2026 Stryker attack showed what happens when an attacker reaches it: the platform's own wipe command erased tens of thousands of devices in hours. This guide covers three things: making enrolment trustworthy, using compliance as a reliable access signal, and treating the MDM console, its administrators, its API integrations and its certificates as tier-0 assets. It is vendor-neutral and applies to Microsoft Intune, Jamf, Workspace ONE, Ivanti and similar unified endpoint management tools. For day-to-day mobile controls such as app protection and BYOD privacy, see the Mobile Device Security Guide.
Key takeaways
- The MDM is tier-0 infrastructure. Whoever controls it can run code on, reconfigure or destroy every managed device, so protect its admins like domain or tenant administrators.
- Destructive actions need a second person. Require multi-admin approval for wipes, scripts, app deployments and role changes, as CISA urged after the Stryker incident.
- Enrolment is where trust is created. Use Apple Automated Device Enrollment, Android zero-touch and equivalent hardware-backed enrolment so corporate devices cannot be set up unmanaged or by an attacker.
- The MDM's own non-human identities are attack paths. API apps, service principals, enrolment tokens and the Apple push certificate need owners, least privilege, monitoring and renewal plans.
- Compliance only works if non-assessed means non-compliant. Devices without a policy, stale check-ins and long grace periods quietly turn compliance into a rubber stamp.
Why the MDM is a tier-0 system
MDM was designed to give a small team total control over a large fleet. The same capabilities are exactly what an attacker wants, and MITRE ATT&CK lists them under T1072 Software Deployment Tools: adversaries use centralised management software to execute commands and move laterally across an enterprise. NIST SP 800-124 Rev. 2, the US guideline for managing mobile devices in the enterprise, covers the whole device lifecycle including centralised management; this guide focuses on protecting that central point.
| MDM capability | What an attacker does with it | ATT&CK | Primary control |
|---|---|---|---|
| Wipe, retire, delete | Mass destruction of devices and data | T1485 Data Destruction, T1561 Disk Wipe | Multi-admin approval, scoped roles, rate alerts |
| App and script deployment | Push malware or a remote access tool to every device | T1072 Software Deployment Tools | Approval for apps and scripts, signed packages, change review |
| Configuration and compliance policies | Disable encryption, firewall or defender; mark devices compliant | T1072 Software Deployment Tools | Approval for configuration changes, drift detection |
| Role-based access control | Create a new top-tier admin to persist | T1098.003 Additional Cloud Roles | Approval for RBAC changes, alerts on new privileged roles |
| Admin and API accounts | Log in with stolen credentials or tokens | T1078.004 Cloud Accounts | Phishing-resistant MFA, Conditional Access, short-lived tokens |
| Internet-facing MDM servers | Exploit the server itself to reach device data and policies | T1190 Exploit Public-Facing Application | Rapid patching, reduced exposure, KEV monitoring |
The on-premises risk is real. CISA's alert on Ivanti Endpoint Manager Mobile CVE-2023-35078 described unauthenticated access to API paths that exposed names, phone numbers and device details, and allowed an attacker to change configurations, including installing software and altering security profiles on enrolled devices. CVE-2023-35078 and CVE-2023-35081, and later CVE-2025-4427 and CVE-2025-4428 in the same product, are all in the CISA Known Exploited Vulnerabilities catalog. A self-hosted MDM belongs in the same patch tier as VPNs and firewalls; see the Edge Device Security Guide.
Enrolment that creates trust
Enrolment decides which devices the organisation trusts and how much control it has over them. The goal is that a corporate device cannot be set up without management, and that a personal device can only enrol with a verified identity into a limited mode.
| Method | Ownership | Security properties | Watch for |
|---|---|---|---|
| Apple Automated Device Enrollment (via Apple Business or Apple School Manager) | Corporate | Devices are supervised automatically (iOS 13, iPadOS 13.1, macOS 10.14.4 or later); enrolment is applied during setup; supervision unlocks extra restrictions and, on iPhone and iPad, network monitoring and device location | Access to the Apple Business account and the server token that links it to your MDM |
| Android zero-touch | Corporate | Devices bought through an authorised reseller check for an assigned configuration on first boot and after factory reset, then provision as fully managed | Portal admin roles; devices without a configuration boot unmanaged; deregistration on resale |
| Windows Autopilot | Corporate | Hardware hash registration links the device to your tenant before shipping | Who can import hashes and assign profiles |
| Apple Configurator supervision | Corporate, existing stock | Supervises iPhone and iPad manually; the device is erased in the process | Physical handling; use for devices outside ADE |
| Account-driven user enrolment (Apple) or Android work profile | Personal (BYOD) | Work data separated; the organisation cannot wipe the whole device | Users enrolling personal phones into full device management by mistake |
- Protect the enrolment portals: Apple Business, the Android zero-touch portal and Autopilot are privileged consoles. Google recommends 2-step verification on zero-touch admin accounts; the portal has Owner, Admin, Manager, Assigner and Viewer roles, and keeps audit logs for up to one year. Give most staff Assigner or Viewer.
- Require MFA at enrolment: so a stolen password cannot register an attacker's device as trusted, and cap devices per user.
- Block personal devices from full enrolment: use platform enrolment restrictions so personally owned devices can only use user enrolment or a work profile. In the Stryker incident, employees who had enrolled personal devices lost personal data when those devices were wiped.
- Close the reset loophole: with ADE and zero-touch, a factory reset returns the device to managed setup. Combine this with Activation Lock management on Apple devices so a stolen, reset device stays bound to the organisation.
- Lifecycle exits: release devices from Apple Business and deregister them in zero-touch before resale, or the next owner's device will try to enrol into your tenant.
Compliance as an access signal
Compliance policies turn device state into a yes or no that the identity provider uses for access. The Device Trust and Posture Guide covers posture signals in depth; the MDM-specific design points are below.
- Unassessed means non-compliant: in Intune, change "Mark devices with no compliance policy assigned as" from Compliant to Not compliant. Check the equivalent default in other platforms.
- Stale devices fall out: shorten the compliance validity period so devices that stop checking in lose access within days, not a month.
- Grace periods with an end: mark non-compliant immediately for security settings such as encryption and OS version, and allow grace only for user actions such as a passcode change.
- Feed threat signals: integrate mobile threat defence or endpoint detection risk levels so a compromised device becomes non-compliant. See the Mobile Threat Defense Buyer's Guide.
- Enforce at the identity provider: compliance is only a control if Conditional Access or equivalent requires it for corporate resources. See the Conditional Access Guide.
Remember that compliance is reported by the device and the MDM. An attacker with MDM admin rights can edit compliance policies to mark compromised devices as compliant, which is another reason configuration and compliance changes need approval.
Securing the MDM itself
Administrator access
- Dedicated admin accounts: separate from email and browsing accounts, with phishing-resistant MFA and Conditional Access that requires a compliant privileged access workstation. CISA's March 2026 alert asked for phishing-resistant MFA and Conditional Access for privileged accounts and least-privilege role design.
- No standing top-tier roles: in the Stryker incident the attacker compromised an administrator and then created a new Global Administrator account for tenant-wide control. Make tenant-wide roles just in time and alert on every new assignment.
- Scoped roles: give helpdesk, app packaging and policy teams only the actions and device groups they support. In Intune that means custom roles and scope tags; the Microsoft Intune Guide covers the role and scope tag pitfalls. Jamf, Workspace ONE and others have equivalent role and site models; see the Jamf Guide.
- Break-glass: keep two emergency accounts excluded from approval workflows only where unavoidable, monitored and tested. See the Break-Glass Account Guide.
Multi-admin approval for destructive and code-running actions
CISA described a second administrator's approval for high-impact actions as the control most directly tied to the Stryker incident, and asked organisations to apply the same protections to other endpoint management tools, not just Intune. In Intune this is Multi Admin Approval, which can protect apps, scripts, device actions such as wipe, retire and delete, configuration, compliance and role changes. Where a platform has no native approval workflow, put the high-impact actions behind a change process, restrict them to a very small group and alert on every use.
| Action | Impact | Approval | Alert |
|---|---|---|---|
| Full wipe | Erases the device, including personal data on fully managed personal devices | Always | Every event; page on more than a handful per hour |
| Retire or selective wipe | Removes corporate data and management | Bulk actions | Bulk events |
| Delete device record | Removes the device from inventory and compliance | Bulk actions | Every event |
| Script or app deployment | Code execution on every targeted device | Always | New or changed assignments to large groups |
| Compliance and configuration change | Can weaken or disable protections fleet-wide | Always | Changes outside change windows |
| Role or permission change | Creates new privileged administrators | Always | Every event |
| Remote lock or passcode reset | Denial of service to a user | Not usually | Bulk events |
Certificates, tokens and integrations
- Apple push certificate: the MDM push certificate is valid for 365 days with a 30-day grace period after expiry, and must be renewed with the same Apple account that created it. Microsoft advises a company email address monitored by more than one person, never a personal Apple account. Losing that account means losing the ability to renew and therefore to manage Apple devices.
- Enrolment program tokens: the server tokens linking Apple Business and other programs to the MDM expire and grant device assignment. Track them in the same calendar as certificates. See the Machine Identity, PKI and Certificate Lifecycle Guide.
- API apps and service principals: reporting, ticketing, inventory and migration tools often hold device management permissions through an API. Give each one an owner, the narrowest permission (read rather than write, no privileged operations unless required), certificate or federated credentials rather than long-lived secrets, and remove them when the project ends. Any app excluded from approval workflows is an approval bypass if compromised. See the Azure RBAC and Service Principals Guide.
- Connectors and on-premises servers: certificate connectors, directory connectors and self-hosted MDM servers need patching to the same deadlines as other internet-facing systems, with the KEV catalog as a trigger.
- Configuration backup: export policies, profiles and role definitions regularly so you can rebuild after malicious changes or deletion.
Where identity fits
An MDM is an identity system in disguise. It issues device certificates, stores enrolment tokens, holds the Apple push certificate and grants API access to other tools, and each of these is a non-human identity that can be stolen or over-scoped. The human side matters as much: a single phished admin session is enough to reach every device. Treat MDM admin roles as privileged access in your PAM programme and the MDM's API apps as non-human identities in your inventory. See the Privileged Access Management Guide and the OWASP NHI Top 10 NHI5 Overprivileged NHI Guide.
Choosing controls for your size and platform
- Small organisations (one or two admins): cloud MDM rather than self-hosted, corporate devices through ADE or zero-touch, phishing-resistant MFA for the admins and a shared, monitored mailbox for vendor accounts. Multi-admin approval needs at least two approvers, so an outsourced IT partner can act as the second approver if a contract defines it.
- Mid-sized organisations: custom roles for helpdesk and packaging, approval for wipes, scripts and roles, a SIEM feed of MDM audit logs and a quarterly API app review.
- Large or federated organisations: scope by region or subsidiary so no regional admin can act globally, separate tenants or instances for high-risk business units where justified, a dedicated MDM platform owner and automated drift detection against exported baselines.
- Self-hosted versus cloud: self-hosted MDM gives data location control but adds an internet-facing server that has been repeatedly exploited. If you keep it, restrict admin interfaces to internal networks and patch exploited vulnerabilities within days.
How to implement
- Inventory privilege: export every admin role assignment, every tenant-level role holder with MDM power, every API app or service principal with device management permissions, and every certificate and token with its expiry.
- Fix admin identity: dedicated accounts, phishing-resistant MFA, Conditional Access requiring privileged workstations, and just-in-time activation for tenant-wide roles.
- Scope roles: build custom roles and scope, move daily administrators off the top-tier roles, and test that each role can do its job and nothing more.
- Turn on approval in stages: device actions and scripts first, then apps, configuration and compliance, then role changes last, after the approver group's own permissions are in place. Pilot with a small admin group for two weeks.
- Harden enrolment: connect Apple Business, zero-touch and Autopilot; restrict personal devices to user enrolment or work profiles; require MFA at enrolment.
- Tighten compliance: set unassessed devices to non-compliant and shorten validity in report-only Conditional Access first, review the impact, then enforce.
- Rollback plan: keep previous policy versions assigned to a rollback group, keep at least two approvers available at all times, and document how to remove an approval policy in an emergency.
- Monitor: stream MDM audit logs to the SIEM and implement the detections below before calling the work complete.
Detection and measurement
- Audit events to alert on: new privileged role assignments, bulk wipe, retire or delete actions, new scripts or apps assigned to large groups, compliance policy edits, approval policy changes and new API app permissions.
- Sign-in signals: MDM admin sign-ins from new locations or non-compliant devices, and admin accounts used outside working hours.
- Rate thresholds: a burst of device actions in minutes is rarely legitimate; in the Stryker incident wipes ran across a window of roughly three hours.
- Expiry monitoring: alerts 60 and 30 days before push certificates and enrolment tokens expire.
| Metric | Target |
|---|---|
| Admins with standing tenant-wide roles | Only break-glass accounts |
| Destructive and code-running actions protected by approval | 100 per cent |
| Corporate devices enrolled through ADE, zero-touch or Autopilot | 95 per cent or more |
| API apps with an owner reviewed in the last 90 days | 100 per cent |
| Devices compliant within policy validity period | Tracked weekly; unassessed devices at zero |
| Days to patch KEV-listed vulnerabilities in self-hosted MDM | Within the KEV due date, ideally days |
Common failure patterns
- One account can wipe the fleet: standing top-tier roles and no approval for wipes. This is the core lesson of the Stryker Intune wiper attack.
- Personal devices fully enrolled: BYOD phones enrolled as fully managed devices, so a wipe destroys personal data and raises legal and employee relations problems.
- Vendor accounts tied to one person: the Apple push certificate or Apple Business account belongs to an employee who leaves, and renewal fails.
- Compliant by default: devices with no policy or no recent check-in still pass device compliance checks.
- Forgotten API apps: an old migration or reporting tool keeps write permissions to device management and is excluded from approval.
- Self-hosted MDM patched like an internal app: exploited vulnerabilities such as CVE-2023-35078 are fixed weeks after release rather than days.
Lessons from incidents
On 11 March 2026 an attacker, claimed by the Handala persona that the US Justice Department reportedly links to Iran's Ministry of Intelligence and Security, used Stryker's Microsoft Intune tenant to wipe enrolled devices; a source cited by BleepingComputer put the figure at nearly 80,000 devices in about three hours. The initial compromise method has not been published. What is known points to the controls above: an administrator was compromised, a new Global Administrator was created, and the platform's legitimate wipe command did the damage. No malware needed to spread. CISA's response, a week later, was to harden endpoint management systems in general: least-privilege roles, phishing-resistant MFA with Conditional Access for privileged accounts, and a second administrator's approval for wipes and other high-impact changes.
Framework mapping
| Framework | Reference | Relevance |
|---|---|---|
| NIST SP 800-53 Rev. 5 | AC-19 Access Control for Mobile Devices, CM-8 System Component Inventory | Enrolment, inventory and device restrictions |
| NIST SP 800-53 Rev. 5 | AC-2 Account Management, AC-6 Least Privilege, CM-5 Access Restrictions for Change, CM-3 Configuration Change Control | Scoped MDM roles, approval for changes |
| NIST SP 800-53 Rev. 5 | AU-6, AU-12, IR-4, CP-9, MP-6 | Audit logging, incident handling, configuration backup, sanitisation by wipe |
| NIST CSF 2.0 | ID.AM-01, PR.AA-05, PR.PS-01, DE.CM-09, RS.MI-01 | Device inventory, least privilege and separation of duties, configuration management, monitoring, containment |
| ISO/IEC 27001:2022 Annex A | 8.1 User endpoint devices, 8.2 Privileged access rights, 8.32 Change management | Device policy, admin control and approved change |
| MITRE ATT&CK | T1072, T1078.004, T1098.003, T1190, T1485, T1561 | See the capability table above |
Practitioner checklist
- Classify the MDM, its enrolment portals and its admin accounts as tier-0.
- Remove standing top-tier roles and alert on every new privileged assignment.
- Require a second administrator's approval for wipes, scripts, apps, configuration, compliance and role changes.
- Enrol corporate devices only through ADE, zero-touch or Autopilot.
- Restrict personal devices to user enrolment or work profiles.
- Set unassessed devices to non-compliant and shorten the compliance validity period.
- Move the Apple push certificate and program accounts to monitored shared company addresses and track expiry.
- Review every API app with device management permissions every 90 days.
- Patch self-hosted MDM servers against KEV-listed vulnerabilities within days.
- Stream MDM audit logs to the SIEM with bulk action alerts.
Standards and references
- NIST: SP 800-124 Rev. 2, Guidelines for Managing the Security of Mobile Devices in the Enterprise
- CISA: Ivanti Releases Security Updates for Endpoint Manager Mobile (EPMM) CVE-2023-35078
- CISA: Known Exploited Vulnerabilities Catalog
- Apple: About Apple device supervision
- Google: Zero-touch enrollment for IT admins
- Microsoft: Get an Apple MDM push certificate
- MITRE ATT&CK: Software Deployment Tools
- Redmond: CISA and Microsoft outline Intune safeguards
Related NHI Mgmt Group resources: Microsoft Intune Guide · Mobile Device Security Guide · RMM Tool Security Guide · UEM Buyer's Guide · Browser Security Hardening Guide