Enabling multi-factor authentication is now one of the fundamentals of identity security. But one question is often less visible: what happens after successful authentication?
Once logged in, the user obviously does not enter their password and does not validate an MFA for each action. Microsoft Entra ID and applications rely on tokens to maintain the session and allow access to authorized resources. These tokens themselves become particularly interesting elements for an attacker.
This is precisely one of the problems addressed by Token Protection, a feature of Microsoft Entra Conditional Access designed to reduce token replay attacks. Microsoft defines it as session control that ensures that only device-bound session tokens are accepted in supported scenarios.
Let’s imagine a relatively simple scenario.
A user authenticates normally on his computer. His identity and, if applicable, his second authentication factor are validated. Once the authentication is complete, Microsoft Entra delivers the necessary elements to maintain this session. If an attacker succeeds in stealing an exploitable token, the objective may then be to reuse this session from another device.
The problem is important: the initial authentication may well have been legitimate. Stealing a session token can allow an attacker to try to bypass a new password entry or MFA challenge
In other words: User → Authentication + MFA → Token → Theft of the token → Replay from another device
MFA continues to be of major importance, of course. But it mainly protects authentication. Token Protection adds protection to what happens after authentication, at the session level.
How does Token Protection work?
The principle is actually quite elegant: make the token dependent on the device to which it was issued. When a compatible device is registered with Microsoft Entra, a Primary Refresh Token (PRT) is issued and cryptographically linked to that device.
When Token Protection is enforced, Microsoft Entra verifies that supported applications use device-bound session tokens when they try to access protected resources.
The operation can be summarized as follows:

Microsoft clarifies that endpoint binding means that even when a malicious actor manages to steal the token, it cannot be used from another device in the scenarios covered. This is very different from a simple check on whether the user has the correct ID or whether they have passed their MFA. The device enters the chain of trust itself.
Particularly interesting protection for Microsoft 365
The feature becomes particularly relevant when compared to the resources to which a corporate identity can give access. For native apps, Microsoft currently indicates support for Token Protection for:
- Exchange Online ;
- SharePoint Online ;
- Microsoft Teams ;
- and, on Windows, Azure Virtual Desktop and Windows 365.
Microsoft’s Windows guide also lists many compatible applications, including Outlook, OneDrive, Word, Excel, PowerPoint, Microsoft Teams, Microsoft Copilot, Microsoft Loop, OneNote, Power BI Desktop, Visual Studio, and Visual Studio Code in Microsoft-documented configurations.
This is especially interesting since these apps are at the heart of many Microsoft 365 environments.
Why implement Token Protection?
1. Because MFA doesn’t solve everything
This is probably the key point to remember. Implementing MFA can significantly strengthen the authentication step. But an organization must also think about the protection of the session obtained after this authentication.
Token Protection therefore provides a complementary layer by reducing the possibility of replaying a token linked to a device elsewhere.
The aim is not to replace MFA, but to complete the safety chain.
2. Because identity is now a real security perimeter
In Microsoft 365, an identity potentially provides access to multiple business resources.
Security can therefore no longer be limited to couples: User + Password
We must gradually reason in terms of: Identity + authentication + device + session + resource.
Token Protection fits particularly well into this Zero Trust logic: just because a session has a valid token doesn’t have to be enough. The system also checks the relationship between the session and the device in supported scenarios.
3. Because a stolen token must become as unusable as possible
This is probably the best way to understand the purpose of this technology.
Token Protection does not mean that it is useless to protect workstations against token theft. Microsoft recommends a strategy that combines several mechanisms to Reduce the risk, Detect and mitigate theft, then Protect against replay.
Microsoft cites Microsoft Defender for Endpoint and Microsoft Intune in measures to harden endpoints against token exfiltration methods.
It is therefore a real logic of defence in depth:
Prevent theft → detect compromise → limit the use of the stolen token
Deploying Token Protection yes but…
This feature is powerful, but it would be dangerous to conclude that it is enough to activate a Conditional Access policy for the entire organization. Of course, there are prerequisites and especially compatibility constraints.
For example, Microsoft says that a Microsoft Entra ID P1 license is required to use this feature. The Windows guide also recommends that users have compatible and up-to-date client versions before the policy is applied, as older or unsupported clients may be blocked.
At the time of writing, Microsoft says that Token Protection for native apps is generally available on Windows, iOS/iPadOS, and macOS. For web apps, support remains in preview and is for specific apps and configurations accessing Azure Resource Manager.
This is an important point: Token Protection is not currently a universal protection that can be applied indiscriminately to all applications, browsers and devices.
For the preview of web apps, Microsoft recommends starting with Token Protection for native apps with a pilot group before experimenting with this web protection.
In a modern Microsoft 365 environment, a consistent approach can combine Conditional Access, Token Protection, endpoint management and compliance with Microsoft Intune, and endpoint protection with Microsoft Defender for Endpoint. Microsoft positions these different technologies as capabilities that help reduce the risk of token theft and replay.
Of course it still deserves a little investigation but between us … do we have a choice?
