Rollout & operations · Account security

Moving from SMS one-time codes to passkeys, step by step

5 min read

Key points

  • SMS one-time codes can be relayed through fake sites and stolen through SIM swaps, and NIST treats authentication over the phone network as a restricted authenticator.
  • Offer passkey creation right after a successful SMS check, then switch people who have a passkey to passkey sign-in.
  • Don't drop SMS at once: decide where it remains, for people without passkeys and for recovery, and narrow that over time.

To move from SMS one-time codes to passkeys, offer passkey creation right after someone passes an SMS check, and switch sign-in to passkeys for each person who registers one. Don't turn SMS off right away. Decide where it stays, for sign-in by people without a passkey and for account recovery, and narrow that over time. SMS codes can be typed into a fake site and stolen through SIM swaps, which is why NIST classifies authentication over the phone network as a restricted authenticator.

What's wrong with SMS codes

They can be phished

People type the six digits they receive into whatever page is in front of them. If that page is a fake, they can't tell, and the fake site simply relays the code to the real one. NIST SP 800-63B-4 states plainly that OTP authentication and out-of-band authentication are not phishing-resistant.

SIM swaps and network weaknesses

The US Cybersecurity and Infrastructure Security Agency (CISA) lists SIM swaps (an attacker convinces a carrier to move the victim's number to the attacker's SIM) and exploitation of the SS7 signaling protocol as attacks that SMS and voice MFA are exposed to. CISA recommends using SMS MFA only as a last resort, and as a temporary measure while moving to stronger MFA.

NIST's "restricted" status

NIST SP 800-63B-4 names out-of-band authentication over the public switched telephone network (PSTN) as the only restricted authenticator at the time of publication. Organizations that use it must offer at least one unrestricted alternative, tell users about the risks, and have a migration plan in case it becomes unacceptable. NIST also recommends checking for risk indicators such as a device swap, SIM change or number porting before sending a code.

What passkeys change

With a passkey, the device signs a challenge bound to the origin and RP ID the browser verified. A fake site can't obtain a signature for the real one, and there is no code to relay. Passkeys don't depend on a phone number, so SIM swaps don't affect them. The same CISA fact sheet describes FIDO/WebAuthn as the only widely available phishing-resistant authentication.

It's also less work for people: instead of waiting for a text and copying digits, they use their fingerprint, face or screen lock.

The migration, in order

1. Take stock

  • Every screen that uses SMS codes (sign-in, password reset, confirming sensitive actions, changing the phone number)
  • Which devices your users sign in from (mostly phones, or mostly computers)
  • How often people contact support because codes arrive late or not at all

2. Offer a passkey right after an SMS check

Your first entry point is the moment after someone signs in with an SMS code. They have just proven who they are, and "next time, sign in without waiting for a code" is an easy offer to accept. The FIDO Alliance's design guidelines include a pattern that introduces passkeys right after SMS OTP verification.

Prompting users to create a passkey after sign-in covers how to design that screen.

3. Make passkeys the default for people who have one

For people who have registered a passkey, show passkey sign-in first from then on. With autofill (conditional UI), the passkey appears as a suggestion as soon as they tap the username field. At this stage, keep a link to switch to SMS for devices where passkeys aren't available.

4. Narrow where SMS remains

As passkey adoption grows, revisit each place SMS is used.

Situation Example of a gradual approach
Sign-in for people without a passkey Keep it for now, and keep encouraging passkeys
Sign-in for people with a passkey Passkey by default, SMS only as a fallback when the passkey can't be used
Recovery after a lost device Don't rely on SMS alone; combine it with other checks such as the registered email address or support staff verification
Changing the phone number Require re-authentication with a passkey

The last row matters most. If an attacker can still get in through SMS, an account protected by a passkey is only as strong as SMS. Whether you can stop SMS sign-in for passkey users depends on how good your recovery paths are.

5. Change the default for new accounts

Encourage new users to create a passkey from the start. If you require a phone number, this is a good moment to ask whether you really need it, and whether it's for something other than authentication, like delivery notices.

Recovery paths

If SMS remains as an identity check, recovery tends to become the weakest way in. For people who lose their device, consider combining:

  • A second passkey (another device or a security key) registered in advance
  • A check against the registered email address in addition to SMS
  • Support staff or document checks where large transactions are involved
  • A notification to the user after any recovery

See Designing passkey account recovery and Helping users who lost or replaced their device for more.

Implementing this with SealGate

SealGate is an API for passkey registration, sign-in, revocation and recovery. It does not send SMS messages or verify identity. Keep your existing SMS code flow, and on the signed-in page after a successful SMS check, call sg.register() from the browser SDK to register a passkey. For people who lose their device, verify their identity on your side first, then call /v1/recovery/start to issue a recovery token valid for 15 minutes, and have them register a passkey on the new device. The docs describe the flow.

FAQ

Should we stop SMS right away?

No. While some users haven't registered a passkey or use devices that can't, keep SMS as a sign-in option. Narrow where it applies as adoption grows.

Are authenticator app codes (TOTP) safe?

They aren't affected by SIM swaps, but they can be typed into a fake site just like SMS codes. NIST treats OTPs as not phishing-resistant.

Our account ID is a phone number. Can we use passkeys?

Yes. Passkeys don't depend on a phone number. You only link the passkey to your own user ID.

Sources

Related posts

Back to the blog