Every login is a small bet. The service bets that the person typing is the person who owns the account. A password alone is a thin basis for that bet, and the federal guidance says so plainly.

Close-up of a laptop beside reading glasses and a folded note on a worn wooden desk, lit softly from a side window, representing a password login step.
A laptop on a desk beside reading glasses and a folded note, the kind of setting where a password is typed and nothing else is checked.

NIST's Digital Identity Guidelines treat memorized secrets, meaning passwords, as the most common authenticator. They also say a password used alone fits only a low-security need. Anything with real risk, including release of personal information, calls for multi-factor authentication. NIST cites Executive Order 13681 for that point. So the question is not whether to add a second factor. It is which one.

Which version of the rules applies

Older write-ups of this topic point to SP 800-63B from the 800-63-3 series. That revision was superseded by SP 800-63-4 as of August 1, 2025. If you read a summary, check its date. Some articles still describe the older rules as current.

The current text is on NIST's own site, at SP 800-63B-4. It defines requirements at three authentication assurance levels, called AALs. Authentication here means a verifier checking that a claimant controls authenticators tied to a subscriber account. The guidelines recognize nine authenticator types. NIST also keeps an FAQ for practitioners, though it is written for agencies and developers, not for consumers.

SMS codes: familiar, and restricted

A text message code is a possession-based, out-of-band method. You prove you hold a phone. It is easy to roll out and easy to use, which is why it is everywhere.

It is also weak in known ways. Under SP 800-63B-4, authentication over the public switched telephone network, which includes SMS, is classed as a restricted authenticator. Restricted means a service that offers it should accept extra risk and have mitigations. The risks come from the phone network itself, plus SIM changes and number porting. An attacker who talks a carrier into moving your number can receive your codes. A fake login page can also relay a code in real time. Rate-limiting failed guesses does not stop either attack.

A quiet home desk at dusk with a closed laptop, a face-down phone and a pen beside a folded paper notice, lit by soft side window light.
A face-down phone at dusk, the device that receives text codes and the one a SIM swap would target.

Good services add protections around SMS. Practitioner summaries of the guidance describe several:

  • Limit repeated wrong code attempts.
  • Make each code work only once.
  • Never send codes by email or VoIP.
  • Check that the number belongs to a mobile network, not a VoIP or software line.
  • Require a second authentication step before the registered number can be changed.

SMS still beats a password alone. But for new systems, one security trade publication summarizing NIST's position advised considering alternatives.

Phishing resistance and hardware keys

The key property is verifier impersonation resistance. It means a fake site cannot capture your credential and replay it to the real one. One-time codes lack it. A practitioner summary notes that OTP, paired with a password, can reach AAL2 but cannot reach AAL3 for that reason.

That same summary recommends phishing-resistant factors such as FIDO2 or passkeys where real-time phishing is a worry. Hardware security keys fall in that family. The sources I had do not explain how keys work for everyday users, what they cost, or how to recover if you lose one. Check the vendor's own documentation on those points before you buy.

One more design idea is authentication intent. It means you take a deliberate action to approve a login, so nothing is approved by accident.

What to ask and what to avoid

The right method depends on the customer, the channel and the level of risk. It can also change over time. A savings account and a forum login do not need the same lock. Three questions help when a service offers you options:

  1. Does it send codes by SMS, email or an app, and does it check that the number is a real mobile line?
  2. Is the method phishing-resistant, and what happens if your number is changed or ported?
  3. Does it limit repeated wrong code attempts?

Some red flags are clear. Security questions as a second factor are one. NIST withdrew knowledge-based authentication because the answers are often private but not secret, and people reuse them across sites. Codes sent by email or VoIP are another. So is any service that lets someone change your phone number with no second check.

The answers to security questions are often private but not secret, and they get reused across sites.

Gaps remain. The sources here are technical standards and vendor material aimed at organizations. They do not include CFPB, FTC, FDIC or Federal Reserve advice on setting up MFA for bank accounts, or steps for recovering from a SIM swap. For those, go to your financial institution and to those agencies directly.