Explore our new Cybersecurity Terms Glossary, explained by our experts.

View the Glossary View the Glossary
PRIVIA

Cybersecurity Terms Glossary

What Are Conditional Access Policies?

Conditional access policies evaluate who is signing in, from what device, to what application, and under what risk, then grant, block, or demand a stronger factor.

Illustration for the cybersecurity term Conditional Access Policies
Home / Blog / Cybersecurity Terms / What Are Conditional Access Policies?
17 November 2026

What Are Conditional Access Policies?

What Are Conditional Access Policies?

A password check answers one question: does this person know the secret? A conditional access policy asks several more before letting the session proceed. Which account is this, which application are they opening, is the device enrolled and compliant, where is the request coming from, and does the sign-in look like the ones that account usually makes? The output is not a yes or no but a decision: allow, block, or allow only after a stronger proof.

Microsoft Entra ID popularised the term and the policy builder most UK organisations will recognise, but the same engine exists as adaptive access in Okta, context-aware access in Google Workspace, and access policies in Ping Identity. The design questions are identical.

Signals, Conditions, Controls

Every policy is an if-then statement built from three parts.

This is attribute-based access control applied at the moment of authentication, and it is the practical mechanism behind most zero trust programmes in Microsoft estates.

The Policies Worth Having First

A small set of policies removes most of the realistic attack surface, and there is a reasonable order to deploy them in.

  1. Require phishing-resistant multi-factor authentication for all administrative roles. Password plus push notification is not enough for Tier 0.
  2. Block legacy authentication protocols. IMAP, POP, and older Exchange clients cannot present a second factor, so they were the standing bypass for every MFA deployment until Microsoft began disabling them by default.
  3. Require multi-factor authentication for all users, with a defined exclusion list that is reviewed monthly.
  4. Require a compliant or managed device for access to sensitive applications, which closes the gap where a valid credential on an unmanaged machine is sufficient.
  5. Require reauthentication and restrict downloads when risk is elevated or the device is unmanaged.

Where Attackers Find the Gaps

Conditional access fails at its edges, not in the middle. The recurring gaps are worth naming.

Exclusions that never expire. Every deployment accumulates exclusions—a service account, an executive who objected, a legacy integration. Each is a permanent hole in the policy, and attackers enumerate them.

Token theft after the check. Policies evaluate at sign-in. An adversary-in-the-middle kit such as Evilginx captures the resulting session token and replays it, having satisfied every condition using the real user. Continuous access evaluation, token protection, and phishing-resistant factors are the answers; a push-based second factor is not.

Flows that do not prompt. The device code flow has been used repeatedly since 2024 by state-linked actors to obtain tokens without ever triggering an interactive policy against the attacker’s own device. Restrict it to the specific client applications that need it.

Location as a security boundary. Trusted-network exclusions assume the corporate IP range is trustworthy. A compromised VPN account or an on-premises foothold makes that assumption false, and an attacker inside the range gets a weaker check than one outside it.

Rolling Out Without Locking Yourself Out

The classic failure is a policy that applies to all users including every administrator, with a condition that no one can currently satisfy. Two or three cloud-only break-glass accounts, excluded from every policy, protected by hardware security keys with credentials split under dual control, and alerting on every single use, are the insurance. Test them quarterly.

Beyond that: use report-only mode first and read the sign-in logs before enforcing, use the what-if evaluation tool against real accounts, and deploy to a pilot ring before the organisation. Keep policies few and readable—twenty overlapping policies with contradictory exclusions is an outage waiting for a Monday. Finally, treat every policy change as a privileged change: alert on creation, modification, and deletion, because quietly adding an exclusion is an effective and under-monitored persistence technique after an administrative account is compromised.

At Privia Security, we review conditional access design and test it against real session-theft techniques as part of our red team services.

FAQ – Frequently Asked Questions

Do Conditional Access Policies Replace Multi-Factor Authentication?

No, they decide when it is demanded and which form counts. A conditional access policy can require any second factor for ordinary access, insist on a phishing-resistant factor for administrative roles, and skip the prompt on a compliant managed device. The authentication method still does the work; conditional access supplies the judgement about how much proof the situation warrants.

What Is a Break-Glass Account?

A cloud-only emergency administrator account excluded from conditional access policies so that a misconfigured or malfunctioning policy cannot lock everyone out of the tenant. Protect it with a hardware security key, split the credential under dual control, exclude it from any policy that could block it, alert on every sign-in, and test it on a schedule so you know it still works.

Why Does Blocking Legacy Authentication Matter So Much?

Legacy protocols such as IMAP, POP, and older Exchange clients predate multi-factor authentication and cannot present a second factor, so they accept a username and password alone. For years this was the standard way to bypass an MFA rollout entirely. Blocking them is usually the highest-value single policy, though it needs an inventory of which applications still depend on them.

Can Conditional Access Stop Session Token Theft?

Only partially, and only with specific features enabled. The policy is evaluated at sign-in, so a token stolen afterwards already satisfied it. Continuous access evaluation shortens the window by revoking sessions on critical events, token protection binds the token to the device it was issued to, and phishing-resistant authentication prevents the interception in the first place.

Related Terms

You May Be Interested In These

What Are Web Skimming and Magecart Attacks?

Cybersecurity Terms

17 November 2026

What Are Web Skimming and Magecart Attacks?

Read More
What Is DNS over TLS?

Cybersecurity Terms

17 November 2026

What Is DNS over TLS?

Read More
What Is Security Posture?

Cybersecurity Terms

17 November 2026

What Is Security Posture?

Read More