Passkeys replace reusable passwords with public-key credentials bound to a legitimate service. A user might approve sign-in using a device PIN or biometric while the browser proves possession of a private key without sending the key to the website. This reduces common phishing and credential-reuse risks, but assurance still depends on the server verifying the challenge, origin, and required authenticator flags. A passkey also does not secure every recovery flow or application session. This guide explains WebAuthn and CTAP, user presence versus verification, synced credentials, and the controls that remain necessary after sign-in.
What changes when an account uses a passkey
Traditional password authentication depends on a shared secret. A user knows it and the service verifies a derived value. If an attacker obtains the password through phishing, reuse, malware, or a weak reset process, the attacker may be able to submit the same secret. One-time codes can reduce risk but may still be relayed through convincing real-time phishing pages.
Passkeys use asymmetric cryptography instead. An authenticator generates a key pair for a relying party, normally identified by a service domain. The private key remains within the authenticator or a supported credential-sync system. The service stores the public key and a credential identifier, not a password-equivalent secret. A sign-in challenge is signed with the private key and verified by the relying party using the public key. This proof-of-possession model helps explain authentication and identity management when reviewing CompTIA Security+ practice questions.
That design changes the attacker’s options. A copied public key cannot produce an authentication signature, and an assertion made for one challenge and service cannot simply be replayed as a valid assertion for another. The credential is scoped to the relying party, so a fake site on a different domain cannot request the legitimate service’s credential as though it were the real origin. The wider architecture is explained through phishing-resistant authentication; the key point is that phishing resistance is a property of the entire correctly configured sign-in path.
How WebAuthn and CTAP work together
FIDO2 combines the Web Authentication (WebAuthn) browser and relying-party model with the Client to Authenticator Protocol (CTAP) used when a client communicates with external authenticators such as security keys. A device’s built-in authenticator can participate through the platform’s interfaces. These layers are related but not interchangeable: a browser API is not the same thing as a hardware key or a server-side identity session.
The W3C Web Authentication Level 3 specification became a Recommendation on August 25, 2026. It describes credential creation, origin and relying-party constraints, signed authenticator data, and what a relying party must verify. That is more specific than a promise that passkeys are “secure by design”: the application still has to check the fields and policy that give the signature meaning.
Registration and sign-in are different security events. During registration, the service sends a fresh challenge, relying-party information, account context, and acceptable cryptographic options. The browser mediates the operation, and an authenticator creates or selects a credential. The service validates the response, stores the public key, and binds the credential to the authorized account. It should prevent a compromised lower-assurance session from silently enrolling a new authenticator.
During authentication, the service sends another unpredictable challenge. The authenticator signs data that includes the challenge-related client context and relying-party information. The service checks the signature, challenge, expected origin, relying-party identifier hash, credential ownership, and required authenticator flags before creating a session. Reusing stale challenges or skipping account binding can undo the security of a strong authenticator.
Server-side libraries reduce implementation risk, but they do not decide which user account should receive a credential, how an administrator can reset access, or whether an authentication event warrants a privileged session. Those remain application and identity architecture responsibilities.
User presence is not user verification
A subtle but important WebAuthn distinction is between user presence (UP) and user verification (UV). UP indicates that the authenticator recorded a user interaction or consent event, such as touching a security key. UV indicates a stronger local verification step, commonly a device PIN, biometric verification, or another authenticator-approved mechanism.
These are separate signed authenticator-data flags. Touching a security key may satisfy UP without satisfying UV. A relying party that requires local verification must request it and check the UV result. It cannot infer that a person was locally verified merely because somebody tapped a key or because the login succeeded. A user-presence signal is not interchangeable with local user verification in the authenticator’s signed output.
| Security check | What it establishes | What it does not establish by itself |
|---|---|---|
| User presence (UP) | The authenticator observed interaction or consent. | That the person completed local PIN or biometric verification. |
| User verification (UV) | An approved local verification requirement was met. | That the web application mapped the credential to the correct account and session. |
| Origin and RP-ID validation | The assertion was produced for the intended service identity. | That account recovery, cookies, endpoint posture, and permissions are secure. |
Organizations should base assurance rules on risk rather than labels. A privileged administrative action might demand UV and recent authentication, whereas a lower-risk consumer interaction could have different requirements. When comparing passkeys with multi-factor authentication, evaluate the authenticator’s real factor properties and fallback routes instead of assuming every passwordless flow is equally strong.
Synced, device-bound, and cross-device credentials
Passkey implementations vary in where credentials are stored and how users regain them. A multi-device or synced credential may be available on several supported devices through a credential provider. A device-bound credential remains on one authenticator, such as a managed security key. Both can use WebAuthn, but they shift operational trust and recovery risk to different places.
Synced passkeys can reduce lockouts when a phone is replaced because credentials may be restored through the user’s credential account. Administrators must then understand that account’s recovery controls, device-management policy, and the provider’s security commitments. Device-bound credentials may support stricter device assurance but require backup authenticators and clear replacement procedures. “Passkey” alone is not a reliable statement about whether a credential is synchronized.
Cross-device authentication is another workflow: a person starts sign-in on a computer and approves using a phone as the authenticator. Supported browser and device proximity or transport mechanisms mediate the assertion; the phone is not simply sending a permanent secret to the computer. Users may still have compatibility issues across browsers, managed devices, and platform policies, so actual client testing matters.
Attestation can sometimes help assess authenticator type, but browsers and credential providers apply different availability and privacy rules. An enterprise should require particular authenticator characteristics only when the resulting assurance justifies enrollment complexity and exclusion of otherwise legitimate users.
Phishing resistance and remaining account risks
Passkeys reduce one major attack path: a fake domain cannot collect a reusable authenticator secret for the legitimate relying party. They do not make a user immune to every phishing or account-takeover technique. Attackers can still target account recovery, steal an existing application session, exploit client-side script vulnerabilities, persuade the user to approve an unintended transaction, or abuse an administrator’s credential-registration privileges.
A browser can authenticate a person through WebAuthn and the application can then issue a session cookie with a separate lifetime and access-control model. That cookie needs secure transport, scope, expiration, revocation, and protection against theft. If an identity provider uses a passkey and later returns an OpenID Connect ID Token to a relying application, the application must still verify token issuer and audience and apply its own authorization. Passkeys do not replace federation protocols or local permissions.
Recovery is part of the actual assurance level. The strongest sign-in method is undermined when a lost-device process allows an attacker to reset the account with easily phished information. A passwordless authentication recovery process should document identity proofing, who may register or remove credentials, notification of changes, and the conditions that require administrator approval or additional proof. Support staff need a process that is workable but not easy to impersonate.
Systems should support planned backup methods where appropriate. People lose phones, change platforms, and need access during hardware failure. A second registered authenticator, recovery through a well-governed identity provider, and audited escalation can reduce lockouts without silently falling back to a weak factor for high-risk users.
The wider zero-trust identity design must treat a completed login as one signal rather than a permanent entitlement. Reauthentication, device health, privilege boundaries, account disablement, and session termination continue to matter after the authenticator has performed its job.
Enterprise deployment and recovery
Start with users and services where phishing resistance materially reduces risk: privileged administrators, access to identity providers, finance and customer systems, and engineers who can change production. Define the relying-party domains, accepted authenticator types, UV policy, fallback methods, credential-registration authority, and support procedures before encouraging every user to enroll.
Domain and account architecture deserve specific testing. WebAuthn credentials are bound to relying-party identifiers and origin rules. A hostname change, multi-tenant domain, or new sign-in portal may affect credential selection. Teams should test real production origins and supported browser combinations instead of assuming credentials registered for one hostname work on every vaguely related domain.
Rollout metrics should reveal the remaining weak path. Count passkey use for real sign-ins, not only credentials registered. Monitor users who still choose passwords or weaker recovery, help-desk enrollment failures, unusual credential additions, and the distribution of supported authenticator types. A program can report thousands of passkeys while the most important accounts still authenticate through easily phished fallback methods.
Security logging should include enrollment, authentication outcomes, authenticator removal, administrative resets, recovery attempts, and session issuance with useful correlation identifiers. It should not store private keys or biometric data. Credentials and session states should be invalidated through defined incident procedures when compromise or device loss occurs.
Testing and measuring effectiveness
Protocol tests should try the paths an attacker would: an unrelated origin, a stale challenge, an unexpected relying-party identifier, a credential associated with a different account, and an assertion without UV when UV is required. The relying party should reject each invalid case without converting it into a fallback flow that is weaker than policy permits.
Operational tests should examine what happens after recovery or revocation. Disable a lost authenticator and then attempt to reuse a previous session, request new credentials, and complete the official recovery path. If the account continues to allow high-risk operations through a session stolen before the reset, disabling the credential alone did not contain the incident.
Evidence-based metrics include adoption of correctly configured phishing-resistant sign-in, the percentage of privileged accounts with approved backup options, anomalous new-credential registration events, session-revocation latency, unsupported client failures, and abuse of recovery processes. These reflect actual security and availability properties better than counting “passwordless users” in a dashboard.
How to apply the model in Security+ SY0-701
The relevant principles are authentication assurance, identity access management, least privilege, and secure recovery. Passkeys provide a useful case study, but the official SY0-701 objectives do not list configuring the complete WebAuthn Level 3 specification as a separately named exam requirement. Reason from the security model, not from a product-specific marketing phrase.
In a practice scenario, identify who holds the private key, what the server validates, which authenticator user-verification check is required, and whether an attacker can exploit session theft or a weaker fallback. That makes the answer defensible even when the scenario uses unfamiliar device or browser terminology.