Product Manager @ Microsoft 👉 Sign up to Entra.News my weekly newsletter on all things Microsoft Entra | Creator of cmd.ms & idPowerToys.com
Microsoft Entra multi-tenant apps can expose multiple customer organizations if a shared vendor credential is compromised. This matters to ISVs, managed service providers (MSPs), and tenant administrators who approve or operate apps across organizations. Merill Fernando’s discussion with Kelvin Tegelaar raises the design and operational trade-offs behind reducing that exposure.
The central issue is the trust model, not a newly announced Entra feature. A vendor can register one multi-tenant application and ask customers to consent to it; each customer that approves it gets a service principal in its own tenant. The credential used to authenticate the app remains with the vendor, so a leak of that shared credential could affect more than one customer.
That is different from a user signing in with a password. The source focuses on application credentials and workload identities: the app authenticates as itself, using a credential such as a secret or certificate. A control aimed at interactive user sign-ins may not address that path in the same way. The result is a cross-tenant risk that deserves review before an organization approves a vendor app.
The source also flags several controls and design choices rather than presenting a single fix. These include app instance property locks, allowed tenants (identified as preview in the source), Conditional Access for workload identities, disabling an application's sign-in, app management policies, and certificate rotation using Microsoft Graph’s operation. Each addresses a different part of the problem; none should be treated as a guarantee that a vendor's shared credential cannot be abused.
This applies most directly to software vendors building services for multiple organizations, MSPs deploying management tools, and Microsoft Entra administrators who grant consent to those tools. It is also relevant when assessing existing enterprise applications: an integration that is no longer needed may still represent an authorized path into the tenant.
The source does not specify a license requirement, tenant type, geographic limit, or general rollout status for the controls it mentions. It describes allowed tenants as preview, but does not provide further eligibility or deployment details. Check the linked Microsoft documentation for the current scope and prerequisites before relying on a particular control.
Administrators should first identify which applications have been consented to, whether the business still needs them, and what permissions they hold. The source does not provide a full inventory procedure or a permissions baseline. It does, however, emphasize that consent places a service principal in the customer tenant while the vendor retains its credential, an important distinction when deciding who controls the risk.
For service providers, the source raises an architectural alternative: use one app per customer rather than a single app shared across all customers. This can reduce the number of customers exposed by one credential, but it comes with an operational trade-off: maintaining separate app instances and credentials. The material does not quantify that effort or prescribe a deployment pattern.
Conditional Access for workload identities is another item to evaluate, but the source explicitly cautions that it cannot cover a vendor's multi-tenant app in the scenario discussed. Review the policy documentation to understand its boundaries; do not assume that enabling a workload identity policy controls every app used by an external vendor.
There is no universal switch here that makes a multi-tenant app safe. An app lock, a tenant restriction, credential policy, or sign-in disablement applies to a particular part of the configuration or response. The right combination depends on the app design and the access the customer has approved.
The source does not give a complete step-by-step configuration for every control, name required licenses, or confirm the rollout status beyond labeling allowed tenants as preview. It also does not provide a full incident-response checklist or quantify the cost of separate app instances. Those details should be verified in current Microsoft documentation and with the vendor before making a production change.
The practical takeaway is to treat vendor consent as an ongoing trust decision. Remove access that is no longer needed, understand whether credentials are shared across customers, and consider how a compromise would be contained. For new services, decide on tenant and application separation early rather than assuming that a single shared identity will remain a low-risk convenience.
Customers can consent to the same vendor app in their tenants while the vendor retains a shared credential. If that credential is compromised, the risk can extend to multiple tenants where the app was authorized.
The source cautions that it cannot cover the vendor's multi-tenant app in the scenario discussed. Review Microsoft's workload identity documentation for the policy's scope and limits.
The source points to Microsoft Entra app management policies for blocking client secrets and capping their lifetimes. Check Microsoft's documentation for configuration details and applicability.
The source labels allowed tenants as preview but gives no broader rollout or eligibility details. Confirm its current status and prerequisites in Microsoft documentation.