---
id: CP-204
slug: conditional-access-risk-based-auth
lang: en
title: Risk-based authentication and conditional access
description: "Why a sign-in is not always handled the same way: location, device, app and risk decide whether to grant, block or verify."
canonical_url: https://startbrain.io/cyberpedia/en/articles/conditional-access-risk-based-auth/
markdown_url: https://startbrain.io/cyberpedia/en/articles/conditional-access-risk-based-auth.md
publisher: Brain (https://startbrain.io)
license: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/)
cite_as: Brain, "Risk-based authentication and conditional access", Cyberpedia, 2026. https://startbrain.io/cyberpedia/en/articles/conditional-access-risk-based-auth/ (CC BY 4.0)
wing: 3
theme: control_access
reflex: cr-027
schema: G3
updated: "2026-11-14"
translations:
  en: https://startbrain.io/cyberpedia/en/articles/conditional-access-risk-based-auth/
  fr: https://startbrain.io/cyberpedia/fr/articles/acces-conditionnel-authentification-risque/
  de: https://startbrain.io/cyberpedia/de/articles/bedingter-zugriff-risikobasierte-authentifizierung/
  es: https://startbrain.io/cyberpedia/es/articles/acceso-condicional-autenticacion-basada-riesgo/
  it: https://startbrain.io/cyberpedia/it/articles/accesso-condizionale-autenticazione-basata-rischio/
  nl: https://startbrain.io/cyberpedia/nl/articles/voorwaardelijke-toegang-risicogebaseerde-authenticatie/
  pl: https://startbrain.io/cyberpedia/pl/articles/dostep-warunkowy-uwierzytelnianie-oparte-na-ryzyku/
  pt: https://startbrain.io/cyberpedia/pt/articles/acesso-condicional-autenticacao-baseada-risco/
  ja: https://startbrain.io/cyberpedia/ja/articles/conditional-access-risk-based-authentication/
sources:
  - "n": 1
    title: What is Conditional Access?
    publisher: Microsoft
    url: https://learn.microsoft.com/en-us/entra/identity/conditional-access/overview
    accessed: "2026-11-14"
  - "n": 2
    title: What are risk detections?
    publisher: Microsoft
    url: https://learn.microsoft.com/en-us/entra/id-protection/concept-identity-protection-risks
    accessed: "2026-11-14"
  - "n": 3
    title: Build a Conditional Access policy
    publisher: Microsoft
    url: https://learn.microsoft.com/en-us/entra/identity/conditional-access/concept-conditional-access-policies
    accessed: "2026-11-14"
  - "n": 4
    title: Token Protection in Microsoft Entra Conditional Access
    publisher: Microsoft
    url: https://learn.microsoft.com/en-us/entra/identity/conditional-access/concept-token-protection
    accessed: "2026-11-14"
  - "n": 5
    title: Zero Trust Architecture
    publisher: NIST
    url: https://csrc.nist.gov/pubs/sp/800/207/final
    published: "2020-08-11"
    accessed: "2026-11-14"
  - "n": 6
    title: NIST Special Publication 800-63B
    publisher: NIST
    url: https://pages.nist.gov/800-63-4/sp800-63b.html
    accessed: "2026-11-14"
  - "n": 7
    title: Zero Trust Maturity Model
    publisher: CISA
    url: https://www.cisa.gov/zero-trust-maturity-model
    accessed: "2026-11-14"
---

# Risk-based authentication and conditional access

Two colleagues sign in to the same mailbox. One enters a password and reaches their email. The other also has to approve a request on a phone, or is refused access from a computer the company does not recognize. The difference comes from conditional access: the company has defined rules that look at the context of each sign-in.

This article describes the principle using Microsoft Entra documentation, then places it within the Zero Trust approach of NIST and CISA. It ends with what an employee can do when an unexpected check appears.

## In 30 seconds

- Where it happens: at every sign-in to a company application or service, for example Microsoft 365 [1].
- What the system does: it combines signals (user, location, device, application, risk) and decides to block access or grant it, sometimes under conditions [1].
- The reflex: an approval request you did not trigger is not approved, and is reported if the reporting feature is turned on. A denial that is reported produces a risk detection [2].
- If it is already done: a multifactor authentication request that the user did not initiate may mean their credentials are compromised [2].

## Straight answers

### Why is my sign-in handled differently from a colleague's?

Conditional Access policies are if-then statements: if a user wants to access a resource, then they must complete an action [1]. Several policies can apply to the same person, and all of them must then be satisfied: if one requires multifactor authentication and another a compliant device, both are needed [3]. What you see therefore depends on the combination of rules that applies to you.

### When do these rules apply?

After first-factor authentication is completed [1]. Conditional access is not intended as a frontline defense against denial-of-service attacks, but it can use signals from these events to decide on access [1].

### What parts of the context are looked at?

Microsoft lists the user or group, IP location, the device, the application, and real-time or calculated risk detection [1]. Administrators can also specify IP ranges for entire countries or regions to block or allow [1].

### What can the system decide?

Blocking access is the most restrictive decision [1]. The other outcome is to grant access while requiring one or more conditions, for example: multifactor authentication, "Require authentication strength", a device marked as compliant, a Microsoft Entra hybrid joined device, an approved client app, an app protection policy, a password change or acceptance of terms of use [1].

### What is a risk signal?

It is a detection produced by Microsoft Entra ID Protection, which Conditional Access policies can use [3]. Policies based on sign-in risk and user risk require this feature, which is part of the Microsoft Entra ID P2 license [1].

## How it works

### Two phases

Microsoft describes an evaluation in two phases. In the first, the system gathers session details, such as network location and device identity [3]. In the second, it identifies the requirements that are not yet satisfied. If a policy provides for blocking, everything stops there and the user is blocked [3]. Otherwise, the user must meet the remaining requirements in this order: multifactor authentication, compliant device, hybrid joined device, approved client app, app protection policy, password change, terms of use, custom controls [3]. Once these requirements are met, session controls apply [3].

### What session controls can change

They can limit the user experience. For example, with Microsoft Defender for Cloud Apps, they can block the download, cut and paste, copying and printing of sensitive documents [3]. Sign-in frequency and whether the session persists after the browser is closed are also set at this level [3].

### The risk signals that trigger a request

The Entra ID Protection documentation lists many detections. A few examples on the sign-in side:

- Atypical travel: two sign-ins from geographically distant locations, at least one of which may be atypical for the user. The algorithm takes into account the time between the two sign-ins and the time it would take to travel. This risk may indicate that a different person is using the same credentials [2].
- Anonymous IP address: a sign-in from an address such as those of the Tor browser or an anonymous VPN, which actors typically use to hide their sign-in information [2].
- Unfamiliar sign-in properties: an IP, ASN, location, device, browser or subnet that differs from the history. A newly created user first goes through a learning period whose minimum duration is five days [2].
- Suspicious MFA authentication approval: unusual properties (ASN, browser, device, GPS location) point to a social engineering or phishing attack. The detection relies on data from the Microsoft Authenticator app [2].
- Anomalous token: an unusual token lifetime, or a token played from an unfamiliar location. Microsoft states that some of the sessions flagged at low or medium risk may be false positives [2].

On the user side, valid credentials found in a known breach always trigger a high-risk detection [2]. An authentication session linked to a malicious reverse proxy (an attack called "Adversary in the Middle") raises the user to high risk [2].

### Protecting the token itself

Token protection is a session control that attempts to reduce token replay attacks by ensuring that only device-bound sign-in session tokens are accepted [4]. When a supported device is registered, a primary refresh token (PRT) is cryptographically bound to it, so a stolen token cannot be used from another device [4]. To limit disruption from incompatible applications or devices, Microsoft recommends starting with a pilot group and creating the policy in report-only mode first [4].

### The Zero Trust framework

NIST (SP 800-207) describes Zero Trust as a set of paradigms that move defenses from static network perimeters to users, assets and resources [5]. No implicit trust is granted to an account or asset based solely on its physical or network location, or on its ownership (company-owned or personally owned device) [5]. Authentication and authorization are discrete functions performed before a session to a resource is established [5]. Microsoft presents conditional access as its Zero Trust policy engine, aligned with three principles: verify explicitly, use least privilege, assume breach [1].

NIST SP 800-63B points the same way: an authentication from an unexpected geolocation or IP address block may trigger additional risk-based checks [6]. These indicators do not replace an authentication factor [6]. For CISA, whose page is marked as archived content, Zero Trust aims to minimize uncertainty in access decisions on each request, treating the network as compromised [7].

## What you do

- If an approval request arrives when you have not started a sign-in, deny it. A multifactor authentication request that the user did not trigger may signal that their credentials are compromised [2].
- Report it as suspicious activity when that option exists: the "User reported suspicious activity" detection appears when a user denies an MFA prompt and reports it as suspicious activity, provided the reporting feature is turned on [2].
- If access is refused although you are the right person, a policy may block based on location or require a particular device, for example a device marked as compliant or managed by the organization [1]. The refusal comes from a rule defined by administrators [1].
- If an extra check appears after travel or from a new device, an unusual location or unusual sign-in properties may trigger a risk detection [2].

## Pitfalls

- The context is not always reliable: the device platform is determined from unverified sources, such as user agent strings, which can be modified [3].
- A risk signal is not proof: some of the sessions flagged by the anomalous token detection at low or medium risk may be false positives [2].
- No signal does not prove there were no attempts: the password spray detection is triggered only when the attacker successfully validates a user's password, and unsuccessful attempts generate no detection [2].
- Without a suitable license, details are missing: customers without Microsoft Entra ID P2 receive a detection titled "Additional risk detected", with no detail about the risk [2].
- Deployment also requires organizational change: CISA (archived page) notes that Zero Trust may require changing an organization's cybersecurity philosophy and culture [7].

## Sources

1. Microsoft, "What is Conditional Access?". https://learn.microsoft.com/en-us/entra/identity/conditional-access/overview (accessed 14 November 2026)
2. Microsoft, "What are risk detections?". https://learn.microsoft.com/en-us/entra/id-protection/concept-identity-protection-risks (accessed 14 November 2026)
3. Microsoft, "Build a Conditional Access policy". https://learn.microsoft.com/en-us/entra/identity/conditional-access/concept-conditional-access-policies (accessed 14 November 2026)
4. Microsoft, "Token Protection in Microsoft Entra Conditional Access". https://learn.microsoft.com/en-us/entra/identity/conditional-access/concept-token-protection (accessed 14 November 2026)
5. NIST, "Zero Trust Architecture", 11 August 2020. https://csrc.nist.gov/pubs/sp/800/207/final (accessed 14 November 2026)
6. NIST, "NIST Special Publication 800-63B". https://pages.nist.gov/800-63-4/sp800-63b.html (accessed 14 November 2026)
7. CISA, "Zero Trust Maturity Model". https://www.cisa.gov/zero-trust-maturity-model (accessed 14 November 2026)

---

Original source: Cyberpedia, the public cybersecurity and AI encyclopedia published by Brain. Canonical version: https://startbrain.io/cyberpedia/en/articles/conditional-access-risk-based-auth/
You may reuse this text freely under the CC BY 4.0 license (https://creativecommons.org/licenses/by/4.0/), provided you credit "Brain, Cyberpedia" with a link to the canonical URL above.
