Passkeys replace a shared secret with a public-key credential. The service stores a public key. The authenticator keeps the corresponding private key and uses it to sign a fresh challenge after the user approves the action.
That changes the security model. There is no password for the server to leak, no reusable secret for the user to type into a convincing fake site, and no password database for an attacker to crack offline.
Passkeys are not magic. Enrollment, device security, synchronization, session handling, and account recovery still determine whether the complete account lifecycle is secure.
How a password login works
A traditional service receives a password and compares a derived value with a stored password verifier. Good implementations use a dedicated password-hashing algorithm with a unique salt. Even then, passwords create persistent problems:
- users reuse them;
- phishing sites can collect them;
- malware and unsafe extensions can steal them;
- weak passwords can be guessed after a database leak;
- recovery channels may be weaker than login;
- SMS or one-time codes can also be phished.
Password managers greatly improve this model, but the service is still accepting a shared secret.
How a passkey login works
During registration:
- The service sends a challenge and credential options.
- The authenticator creates a new key pair scoped to the service.
- The private key remains protected by the authenticator.
- The service stores the public key and credential metadata.
During authentication:
- The service sends a fresh challenge.
- The browser and authenticator verify the relying-party context.
- The user unlocks or approves the authenticator, commonly with a device PIN or biometric.
- The authenticator signs the challenge.
- The service verifies the signature with the public key.
WebAuthn scopes credentials to a relying party and origin. Its Level 3 specification describes discoverable credentials—passkeys—and supports passwordless multi-factor authentication through possession of the private key plus local user verification.[1]
Why passkeys resist phishing
A password can be typed into any text field. A WebAuthn credential is bound to the relying-party identity. A fake domain cannot ask the authenticator to sign in as the real domain.
This blocks the classic real-time phishing proxy from simply forwarding a form and stealing a reusable password or one-time code. The FIDO Alliance describes passkeys as cryptographic key-pair credentials intended for phishing-resistant sign-in.[2]
“Phishing-resistant” does not mean “all social engineering disappears.” An attacker may still:
- trick support into resetting an account;
- steal an authenticated session;
- convince a user to approve an unrelated business action;
- compromise the endpoint or synchronization account;
- exploit an application after authentication.
Passkeys sharply improve authentication, not every layer above and below it.
Synced vs device-bound passkeys
A synced passkey can be made available across a user’s devices through a credential provider. This improves usability and recovery because losing one device does not necessarily lose the credential.
A device-bound credential remains on a particular authenticator, such as a security key or managed device. It can provide stronger device-control and attestation properties for high-assurance environments, but requires careful backup and replacement procedures.
The right choice depends on the threat model:
- consumer services often prioritize cross-device availability;
- workforce deployments may need managed enrollment and device assurance;
- administrators and high-impact roles may justify stricter authenticators;
- regulated environments may need evidence about credential provenance.
Avoid turning a useful distinction into a slogan. A perfectly device-bound credential with a weak help-desk reset process is not a strong account.
Passkeys, biometrics, and privacy
A passkey is not the user’s fingerprint or face. Biometrics commonly unlock a local authenticator. The service receives a cryptographic assertion indicating that user verification occurred; it does not receive the biometric template through WebAuthn.
The device PIN or biometric is therefore a local gate protecting use of the private key. This is why a passkey can combine possession and user verification in one interaction.
What the server stores
The service stores a public key, credential identifier, signature counter or related metadata where applicable, and account association. A stolen public key does not let an attacker create valid signatures.
That removes the “crack the password database” path, but the database still matters. An attacker who can replace registered public keys, alter account mappings, or weaken recovery has another route to takeover. Protect credential-management operations like other privileged account changes.
The recovery problem
Authentication systems are only as strong as their easiest recovery path. If passkey users can reset access using a phishable email plus public personal data, attackers will bypass the passkey.
A mature rollout defines:
- how users register the first credential;
- whether multiple authenticators are encouraged or required;
- what happens after device loss;
- how support verifies identity;
- delays and notifications for high-risk changes;
- how old credentials and sessions are revoked;
- stronger procedures for administrators.
FIDO’s deployment guidance treats phishing resistance as a journey that includes authentication and recovery, not a one-day switch.[3]
A practical migration plan
1. Inventory every authentication path
Include web, mobile, legacy clients, APIs, support, recovery, delegated administration, and machine identities.
2. Offer passkeys alongside existing login
Measure registration, successful use, device changes, fallback rates, and support demand. Explain the user benefit in plain language.
3. Encourage more than one recovery option
Depending on the environment, this may mean synced availability, a second authenticator, an organizational recovery process, or managed-device re-enrollment.
4. Strengthen the fallback
Do not leave password-plus-SMS as a permanent, invisible bypass for every user. Make fallback risk visible and progressively restrict it.
5. Protect credential management
Adding or removing a passkey, changing recovery information, and disabling factors should require recent strong authentication and generate clear notifications.
6. Address sessions
Use secure cookie settings, rotate sessions after authentication, limit high-risk session lifetime, reauthenticate for sensitive changes, and provide session visibility and revocation.
Passkeys vs password managers
They are complementary. Password managers remain essential for services that do not support passkeys, and many can store or coordinate passkeys. A password manager also reduces domain-confusion risk by offering credentials only on matching sites.
For a service designing authentication, passkeys remove more server-side password risk. For a person navigating today’s mixed ecosystem, a reputable password manager remains highly valuable.
The bottom line
Passkeys improve account security because they replace a reusable, phishable secret with an origin-bound cryptographic proof. Their real-world strength depends on the whole lifecycle: trustworthy enrollment, protected devices, careful synchronization, secure sessions, and recovery that does not quietly restore the old risk.
Sources
World Wide Web Consortium, Web Authentication: An API for accessing Public Key Credentials—Level 3. ↩︎
FIDO Alliance, Passkeys: Passwordless Authentication. ↩︎
FIDO Alliance, Passkeys: The Journey to Prevent Phishing Attacks. ↩︎